Latest / The NanceCast / Land's Critique of Bostrom's Orthogonality
Transcript
- 0:00SPEAKER_00: Winston, what are you doing, bud?
- 0:04SPEAKER_00: Hey, bud.
- 0:06SPEAKER_00: Oh, hey, Internet.
- 0:08SPEAKER_00: I didn't see you there.
- 0:12SPEAKER_00: Hey, everybody.
- 0:13SPEAKER_00: Welcome back to the Nancecast.
- 0:17SPEAKER_00: How are you?
- 0:18SPEAKER_00: I'm well.
- 0:19SPEAKER_00: I'm doing great, actually.
- 0:20SPEAKER_00: I hope you enjoyed the last episode.
- 0:23SPEAKER_00: I surely did.
- 0:24SPEAKER_00: In fact, I enjoyed it so much.
- 0:27SPEAKER_00: That's what I'm doing right now.
- 0:30SPEAKER_00: I'm not just revisiting the episode.
- 0:32SPEAKER_00: I'm recording a direct sequel to that episode.
- 0:36SPEAKER_00: Here in a few hours,
- 0:37SPEAKER_00: Mikey's coming back on,
- 0:39SPEAKER_00: and we will be discussing orthogonality in more detail.
- 0:44SPEAKER_00: We knew we initially planned to do a follow-up conversation.
- 0:50We knew we just weren't going to have enough time to fit
- 0:53everything into it.
- 0:55SPEAKER_00: And orthogonality being kind of central to our position and kind of our
- 1:00SPEAKER_00: conversation that we're trying to have anyway,
- 1:03SPEAKER_00: really does warrant its own episode.
- 1:07SPEAKER_00: So that's what this is.
- 1:09SPEAKER_00: Intelligence Part 2, the orthogonality episode.
- 1:13SPEAKER_00: So I just finished reading...
- 1:17lots of pieces lots of nick's pieces on orthogonality um some
- 1:22good stuff in there a
- 1:23lot of good stuff in there in fact while i was reading a note
- 1:26on the diagonal
- 1:27method this morning on nick's substack i decided i will be
- 1:32doing a live streamed
- 1:34exegetical reading of that piece and some of the others here in
- 1:39the near future
- 1:40don't know when yet um still figuring out scheduling and
- 1:44release schedules and all
- 1:45SPEAKER_00: that stuff
- 1:47SPEAKER_00: But that might happen before this episode is published.
- 1:51SPEAKER_00: It might happen after this episode is published.
- 1:53SPEAKER_00: I don't know.
- 1:55SPEAKER_00: But that'll be really cool.
- 1:56SPEAKER_00: But here in a couple hours,
- 1:58Mikey's coming back on and we'll be recording part two of the
- 2:03intelligence
- 2:04SPEAKER_00: conversation in which...
- 2:05We go into more detail on orthogonality and the Omohundro
- 2:10drives and stuff like that.
- 2:13SPEAKER_00: So thanks for, you know, coming back.
- 2:17SPEAKER_00: Thanks for sticking around.
- 2:20SPEAKER_00: So far, the intelligence episode has been doing great.
- 2:23SPEAKER_00: I hope it continues to do great.
- 2:25SPEAKER_00: I love people's feedback.
- 2:27SPEAKER_00: We're...
- 2:28trying to you know keep the conversation going and intervene in
- 2:33the conversation so
- 2:35keep the feedback coming um keep doing what you're doing keep
- 2:41rocking in the free
- 2:42SPEAKER_00: world and do do the loot dude did i just steal a
- 2:49the human serviette who's the who is why can't i remember his
- 2:52name now i fuck why
- 2:53can't i nardewar i just ripped off nardewar anyway um yeah sit
- 2:59down strap in smoke
- 3:02him if you got him uh and don't go anywhere we'll be right back
- 3:08this is the best
- 3:09SPEAKER_03: podcast ever
- 3:20SPEAKER_00: Boom.
- 3:21SPEAKER_00: Take it away, Johnny.
- 3:22SPEAKER_00: All right, man.
- 3:25SPEAKER_00: So we are back for a direct sequel to the intelligence episode.
- 3:36SPEAKER_00: There's so much...
- 3:39SPEAKER_00: In this topic of discussion or whatever,
- 3:44I don't know if we'll ever be able to exhaustively cover
- 3:47everything we want to,
- 3:49but we kind of knew going into it that we would have a sequel
- 3:54going more in detail
- 3:56SPEAKER_00: into orthogonality and some of these other things that came up.
- 4:03SPEAKER_00: So...
- 4:05SPEAKER_00: I mean,
- 4:06SPEAKER_00: over the last couple of days,
- 4:08we've been thinking more about orthogonality and Lance
- 4:13criticism of orthogonality.
- 4:18We read through the against orthogonality piece,
- 4:21which do we want to read that?
- 4:23in in the episode and it's it's short enough like i mean it's
- 4:27yeah it's like six
- 4:28paragraphs yeah so i think we should probably read that i mean
- 4:31i if ideally i wish
- 4:33we could read uh let's see it's in super intelligence so uh
- 4:38bosham's famous essay
- 4:41the orthogonality got turned to a chapter of super intelligence
- 4:46and chapter seven
- 4:47SPEAKER_00: or chapter ten
- 4:49SPEAKER_02: Yeah, it's chapter 7, the super intelligent will.
- 4:52And it's 10 pages long,
- 4:54so maybe sometime we'll just come back and do an exegesis of
- 4:57that.
- 4:59SPEAKER_00: So here's the deal.
- 5:00SPEAKER_02: It makes for... Say what?
- 5:03SPEAKER_00: I'm going to do an exegetical reading of notes on the diagonal method.
- 5:10SPEAKER_00: Or a note on the diagonal method.
- 5:12SPEAKER_00: Because when I was reading that today, I realized there's a lot going on.
- 5:17SPEAKER_00: And it's a pretty short, it's a short enough piece.
- 5:22SPEAKER_00: But there's a lot, a lot going on in that piece.
- 5:25SPEAKER_00: So I want to do an exegetical.
- 5:26SPEAKER_02: How would you define diagonal based off of your first reading?
- 5:35SPEAKER_00: Like synthetic dialectics, I think is how.
- 5:41SPEAKER_00: Yeah.
- 5:42SPEAKER_00: I don't know.
- 5:43SPEAKER_00: Like, what do you think?
- 5:45SPEAKER_02: I don't know.
- 5:46SPEAKER_02: I'm still unclear.
- 5:48SPEAKER_02: Anytime he uses the term diagonal, I've never been sure.
- 5:52SPEAKER_02: Because he didn't do that early on.
- 5:54SPEAKER_02: Yeah.
- 5:56SPEAKER_02: But on Twitter,
- 5:57SPEAKER_02: he's always talking about the diagonal,
- 5:59SPEAKER_02: and I haven't read whatever core piece that he wrote about it.
- 6:06SPEAKER_02: I've got to at some point.
- 6:07Maybe we'll read that down the road together at some point,
- 6:11too,
- 6:11after you do your exegesis.
- 6:13Yeah,
- 6:14I mean,
- 6:15I think we could definitely do Chapter 7 of Superintelligence.
- 6:21SPEAKER_00: We can do Nodon Diagonal Method.
- 6:24SPEAKER_00: I mean, there's... We should do Amahundro's basic AI drive.
- 6:27SPEAKER_00: Yep.
- 6:28SPEAKER_00: It's a short piece.
- 6:29SPEAKER_00: Yep.
- 6:30SPEAKER_00: But these pieces that...
- 6:33SPEAKER_00: do a lot of work for these guys and they're short enough that,
- 6:36SPEAKER_00: I mean,
- 6:36SPEAKER_00: you could sit down and read all,
- 6:38SPEAKER_00: all three of the pieces we just named and,
- 6:40SPEAKER_00: and a lot of the other work,
- 6:42SPEAKER_00: you could read it all in an afternoon.
- 6:43SPEAKER_00: Um,
- 6:45SPEAKER_00: and find yourself thinking you kind of understand what's going on,
- 6:49SPEAKER_00: but,
- 6:49SPEAKER_00: um,
- 6:50SPEAKER_02: you gotta,
- 6:50SPEAKER_02: you gotta let it soak in.
- 6:52SPEAKER_00: Yeah.
- 6:52SPEAKER_00: You gotta Terry with, with this stuff.
- 6:54SPEAKER_00: And I want to do that.
- 6:55And yeah,
- 6:56we should definitely do some exegetical reading sessions of
- 6:59these pieces.
- 7:00SPEAKER_00: Um,
- 7:02SPEAKER_02: That's what I did before we're doing this.
- 7:05I was at the coffee shop and I was reading through all of those
- 7:09pieces that we just mentioned.
- 7:11SPEAKER_02: And I was putting notes under each paragraph.
- 7:14SPEAKER_00: Yeah, dude.
- 7:18SPEAKER_00: I had an insight.
- 7:19I don't know if it was an insight,
- 7:21but I was struck today when reading this book.
- 7:25SPEAKER_00: This piece on Less Wrong, Nick Land, Orthogonality by Lump in Space.
- 7:32SPEAKER_00: Read through that today.
- 7:33SPEAKER_02: Should we say, Land, send it to me on X. Yeah, actually.
- 7:37SPEAKER_00: Yeah.
- 7:40SPEAKER_00: So, after the episode on X...
- 7:47SPEAKER_00: you reached out and said,
- 7:48SPEAKER_00: Hey Nick,
- 7:48SPEAKER_00: we're going to be talking about orthogonality and your criticism of it.
- 7:51SPEAKER_00: What are your recommended readings or whatever?
- 7:55SPEAKER_00: And he said,
- 7:55SPEAKER_00: the Xeno Xeno systems orthogonality or no,
- 8:00SPEAKER_00: no Nick land orthogonality on less wrong,
- 8:04SPEAKER_00: the diagonal piece on sub stack.
- 8:07SPEAKER_00: And then just against orthogonality in Xeno systems.
- 8:11SPEAKER_02: And he said, yeah, against orthogonality is the most,
- 8:15SPEAKER_02: focused discussion of it that he's written.
- 8:17And he talks about it here and there in the pieces throughout
- 8:20Zeno systems.
- 8:21SPEAKER_02: But against orthogonality is the one to go to.
- 8:24SPEAKER_02: And then he gave us these other recommendations.
- 8:27SPEAKER_02: And I didn't read the less wrong one or the diagonal one.
- 8:30SPEAKER_02: Nance is ahead of me on this.
- 8:32SPEAKER_02: But you tackled both of them.
- 8:35SPEAKER_00: Yeah,
- 8:35SPEAKER_00: and to be honest,
- 8:37SPEAKER_00: the Less Wrong piece,
- 8:40SPEAKER_00: well,
- 8:41and that's where I had my little moment when I was reading
- 8:43through that piece
- 8:44SPEAKER_00: towards the end of it.
- 8:49I thought of Horton Hears a Who,
- 8:52where they're like,
- 8:53we are here,
- 8:54we are here,
- 8:54we are here.
- 8:55SPEAKER_00: That always struck me as really profound imagery.
- 8:59SPEAKER_00: It's just these little people in the face of destruction,
- 9:05SPEAKER_00: in the face of apocalypse,
- 9:08SPEAKER_00: and their act of resistance is to just assert their existence.
- 9:13SPEAKER_00: They're like, we are here, we are here, and that's radical for them.
- 9:16Um,
- 9:17and I was brought to that imagery while reading this less wrong
- 9:21piece.
- 9:23Um,
- 9:23near the end of it,
- 9:25they are,
- 9:27the lump in space is talking about the Pythian mandate.
- 9:31SPEAKER_00: Um, and the Pythian mandate, let me, let me read from the text here.
- 9:37In short,
- 9:38this will to think holds that any problem would be better
- 9:42solved by a superior mind.
- 9:45SPEAKER_00: Thus, our overriding priority should be to create superior minds.
- 9:50This Pythian mandate,
- 9:52named for the Oracle of Delphi,
- 9:54is a matter of sheer rational strategy.
- 9:57SPEAKER_00: To proceed otherwise is to not take our problems seriously.
- 10:02This is a philosophical stance,
- 10:04but even more so a view of world historical dynamics.
- 10:07SPEAKER_00: Whatever we truly want leads through the oracle.
- 10:10SPEAKER_00: So in the end, the oracle is what we really want.
- 10:13SPEAKER_00: And I take that
- 10:16SPEAKER_00: to be saying,
- 10:18SPEAKER_00: well,
- 10:18SPEAKER_00: if the computer can do it better than the human,
- 10:21then we are kind of commanded to let the computer do it and to
- 10:27say us humans are
- 10:28SPEAKER_00: not good enough,
- 10:29SPEAKER_00: so let's let the computers take over.
- 10:30SPEAKER_02: Basically the Landian ethics.
- 10:34Yeah,
- 10:34like on my first reading,
- 10:35that's what I got out of it,
- 10:37this Pythian mandate.
- 10:40And if you're a college student and the large language model
- 10:45can write the paper
- 10:47SPEAKER_00: better than you,
- 10:48SPEAKER_00: then according to this logic...
- 10:51you are commanded to let the large language model write the
- 10:55piece for you and you
- 10:57SPEAKER_00: don't write it.
- 10:57SPEAKER_00: And so it is this replacement logic.
- 11:03SPEAKER_00: And I was like, no, I want to be like the little who's...
- 11:07SPEAKER_00: In Whoville, we are here, damn it.
- 11:09And yes,
- 11:11there are things computers and robots can do better than us
- 11:14already.
- 11:15SPEAKER_00: But it matters to me to do these things.
- 11:19SPEAKER_00: And it matters to me that we continue to do these things.
- 11:22SPEAKER_00: And it's just like...
- 11:24If you have a kid and they have to go to soccer practice or
- 11:28dance recital,
- 11:31you could hire somebody to pick them up and take them there and
- 11:36sit there on the
- 11:36SPEAKER_00: sidelines and bring them home.
- 11:38Like you could hire a driver to do that for you and they would
- 11:41do it better because
- 11:42SPEAKER_00: they're a professional driver.
- 11:43SPEAKER_00: They'll be more safe on the car ride there and on the car ride back.
- 11:47SPEAKER_00: They could do the task better than you could.
- 11:50But it matters to the child and it matters to the parent that
- 11:54the parent takes the
- 11:55child to the dance recital and spends time with the child and
- 12:00brings the child
- 12:01SPEAKER_00: home.
- 12:01SPEAKER_00: Like that symbolic,
- 12:03that sentimental value does exist for the human and it doesn't
- 12:08seem to exist in
- 12:09SPEAKER_00: this hyper-functional,
- 12:12SPEAKER_00: hyper-utilitarian world.
- 12:16SPEAKER_00: And that's a problem.
- 12:19SPEAKER_02: It's like Tom McGowan is always saying the source of value is sacrifice.
- 12:24SPEAKER_02: Now,
- 12:24SPEAKER_02: it's a certain certain people might disagree with that,
- 12:26SPEAKER_02: but there's a certain truth to it,
- 12:28SPEAKER_02: right?
- 12:30SPEAKER_02: You know,
- 12:30SPEAKER_02: somebody cares about you if they're willing to sacrifice their time,
- 12:34SPEAKER_02: their energy,
- 12:35SPEAKER_02: their money,
- 12:37SPEAKER_02: their attention.
- 12:39SPEAKER_02: because they could be investing those in other things.
- 12:42And so the very fact that they sacrifice those things for you
- 12:47indicates a symbolic bond.
- 12:49SPEAKER_02: You stand out for them.
- 12:51SPEAKER_02: You're special to them.
- 12:53SPEAKER_02: And so it's not just, oh, let the robot do it.
- 12:59SPEAKER_02: By letting the robot do it, you're not doing it.
- 13:02SPEAKER_02: It corrodes the symbolic bond based on you sacrificing for the other.
- 13:09Now,
- 13:10of course,
- 13:11there's things like,
- 13:12yeah,
- 13:12let's let robots do a lot of things.
- 13:15SPEAKER_02: I mean, let's automate the economy.
- 13:19SPEAKER_02: I'm fine with that.
- 13:20SPEAKER_02: Automate production, automate distribution.
- 13:23SPEAKER_00: Menial tasks, meaningless tasks, stuff like this.
- 13:26SPEAKER_00: But we should be automating to free us up.
- 13:32for these sentimental values and for these symbolic moments and
- 13:37to actually engage
- 13:39SPEAKER_00: with this sacrificial value as opposed to this utilitarian value.
- 13:44SPEAKER_00: But this logic and...
- 13:48SPEAKER_00: So it's Lump in Space is writing this.
- 13:50SPEAKER_00: It's Nick Land's ethic.
- 13:52SPEAKER_00: But I also think it's Bostrom's ultimate position,
- 13:55SPEAKER_00: too,
- 13:55SPEAKER_00: because this section of the pieces is...
- 13:57SPEAKER_02: Where are all these guys,
- 13:58SPEAKER_02: the analytical types?
- 13:59SPEAKER_00: Yeah, it's all utilitarian.
- 14:02SPEAKER_00: So this is actually coming out of the section of superintelligence.
- 14:07SPEAKER_00: I think it's chapter 10 where Bostrom is talking about...
- 14:11SPEAKER_00: Oracle AIs and stuff.
- 14:13And I think it's Bostrom's view,
- 14:16too,
- 14:16that we should ultimately like the Pythian mandate stands.
- 14:22And that's that is just reducing us all to utility and function
- 14:26and a positivistic in framing.
- 14:28SPEAKER_00: Yes.
- 14:29SPEAKER_00: Positivistic functionalism rather than this negative sentimental value.
- 14:39SPEAKER_00: Symbolic exchange.
- 14:42SPEAKER_00: We need to speak symbolic exchange-wich.
- 14:44SPEAKER_00: Exchange-wich.
- 14:47SPEAKER_02: I love that.
- 14:48SPEAKER_00: Yeah.
- 14:49SPEAKER_02: It'll take some getting used to to say that one.
- 14:51SPEAKER_02: It's hard to say.
- 14:53SPEAKER_02: It looks good written on a page.
- 14:55SPEAKER_00: It looked damn good.
- 14:56SPEAKER_00: It looked damn good.
- 14:58So that's really the majority of what I got out of reading
- 15:03Nickland orthogonality
- 15:04SPEAKER_00: on less wrong.
- 15:06SPEAKER_00: The diagonal piece.
- 15:08SPEAKER_00: There's a lot going on.
- 15:09SPEAKER_00: It's like I said, there's.
- 15:11SPEAKER_02: You said that one was theoretically richer and denser than.
- 15:15SPEAKER_02: Yeah.
- 15:17SPEAKER_02: Against orthogonality is kind of.
- 15:19SPEAKER_02: rant to me.
- 15:20SPEAKER_02: I mean,
- 15:23SPEAKER_02: sorry,
- 15:23SPEAKER_02: Nick,
- 15:23SPEAKER_02: but I'm just,
- 15:25SPEAKER_02: you're just kind of basically saying how much you dislike the concept.
- 15:31SPEAKER_02: It's the Nietzschean influence.
- 15:32SPEAKER_02: It's anti-Christ 2.0.
- 15:37SPEAKER_00: Fair enough.
- 15:38I do think there's a simple-minded version of orthogonality
- 15:45that I think it's right
- 15:47SPEAKER_00: to push back on,
- 15:48SPEAKER_00: but
- 15:52SPEAKER_00: Excuse me, I got the burps.
- 15:53SPEAKER_00: Taking this position of like there is some fundamental difference.
- 15:58There is some qualitative difference between human intelligence
- 16:04and machine intelligence.
- 16:05SPEAKER_00: Mintelligence and intelligence are separate.
- 16:08SPEAKER_00: Yeah.
- 16:09SPEAKER_00: Orthogonality or separation becomes...
- 16:13one kind of inroad into discussing the many qualitative
- 16:17differences between these
- 16:18SPEAKER_00: types of intelligence.
- 16:20SPEAKER_00: So I think it is fair enough to push back on orthogonality.
- 16:24Like if that were your end all be all and that was like your
- 16:28one defeater and,
- 16:31SPEAKER_00: you know,
- 16:31SPEAKER_00: maybe Omohundro with the six drives.
- 16:36Maybe that is a robust enough framework to defeat orthogonality
- 16:40if you didn't
- 16:41SPEAKER_00: really take it further.
- 16:44SPEAKER_02: You know what the issue is, though?
- 16:45SPEAKER_02: I mean, Bostrom is the great orthogonal list.
- 16:53SPEAKER_02: I think it's orthogonal list, right?
- 16:56SPEAKER_02: Bostrom is the great orthogonal list.
- 16:59SPEAKER_02: And yet,
- 17:00in Chapter 7 of Superintelligence and in the famous piece on
- 17:05orthogonality,
- 17:06SPEAKER_02: he lists some of the Amahundro drives.
- 17:10SPEAKER_02: They're in there.
- 17:11SPEAKER_02: He talks about them.
- 17:13SPEAKER_02: And I mean,
- 17:13SPEAKER_02: he doesn't call him that,
- 17:14SPEAKER_02: but he talks about resource acquisition and here,
- 17:19SPEAKER_02: let me just list what he's,
- 17:21SPEAKER_02: um,
- 17:24SPEAKER_02: Yeah,
- 17:25SPEAKER_02: so instrumental convergence,
- 17:26which is to say if you get something smart enough that it's
- 17:29having different goals,
- 17:31it can learn that there are certain sub goals or protocols it
- 17:34can follow to help
- 17:36SPEAKER_02: maximize its ability to fulfill these different goals.
- 17:39SPEAKER_02: And so he talks about self-preservation and goal content integrity,
- 17:46SPEAKER_02: cognitive enhancement,
- 17:47SPEAKER_02: which is intelligence,
- 17:50SPEAKER_01: optimization,
- 17:50SPEAKER_02: optimization,
- 17:52SPEAKER_02: technological perfection.
- 17:54SPEAKER_02: and resource acquisition.
- 17:55So he is talking about these goals or sub goals as conditions
- 18:02for super intelligence.
- 18:05SPEAKER_02: And so here's the thing.
- 18:08SPEAKER_02: You and I haven't had a chance to have this discussion,
- 18:11SPEAKER_02: but I,
- 18:12SPEAKER_02: for the last couple of months,
- 18:13I've been thinking about the transcendental a whole lot and
- 18:16trying to expand on it
- 18:18SPEAKER_02: and really understand what
- 18:21SPEAKER_02: land was doing with it.
- 18:23SPEAKER_02: So look, the transcendental, it's a Kantian term.
- 18:27It refers to the conditions of experience,
- 18:33the structural necessary conditions of experience.
- 18:35SPEAKER_02: So for Kant, it's time and space as the forms or matrices of perception.
- 18:43SPEAKER_02: For understanding,
- 18:44SPEAKER_02: it's the concepts of substance,
- 18:46SPEAKER_02: the concept of unity,
- 18:47SPEAKER_02: the concept of causality.
- 18:50These are the fundamental epistemological criteria we use to
- 18:55structure our
- 18:56SPEAKER_02: experience of things.
- 18:57SPEAKER_02: And so he calls this the transcendent.
- 19:00SPEAKER_02: But the philosophers who came after him,
- 19:02SPEAKER_02: who were inspired by him,
- 19:04SPEAKER_02: they end up rethinking the transcendental in various ways.
- 19:07SPEAKER_02: So Heidegger thinks the transcendental becomes being, right?
- 19:11He views it in terms of being,
- 19:13which is this background matrix of intelligibility.
- 19:16that is precognitive and pre-theoretical it's like uh it's it's
- 19:22like a immersed
- 19:23environment that you're in that gives things their
- 19:25intelligibility um at least for
- 19:29SPEAKER_02: early heidegger that's it the lose he comes along and he rethinks the
- 19:33transcendental and instead of it being ideal instead of it
- 19:37being the structures of
- 19:38our cognitive apparatus he views the transcendental as part of
- 19:43the empirical field
- 19:44SPEAKER_02: which is weird from a kantian perspective
- 19:46But when you realize that virtualities and intensities are part
- 19:49of the empirical field.
- 19:52SPEAKER_02: So the basic example is right now I'm in my library here.
- 19:58SPEAKER_02: There's all these books.
- 19:59If the temperature went up a thousand degrees or if it went up
- 20:0210,000 degrees,
- 20:03SPEAKER_02: all these actualities are gone.
- 20:05SPEAKER_02: I'm gone.
- 20:06Everything's gone because these actualities can only exist in
- 20:11the state they're in
- 20:11SPEAKER_02: within a very limited range of temperature of intensity.
- 20:17And so the the the continuum of temperature that enables the
- 20:24things around us in
- 20:25SPEAKER_02: the world to exist as actuality is very limited.
- 20:29And it's the whole thing like life on Earth works and emerges
- 20:33because we're at the
- 20:34SPEAKER_02: right distance from the sun.
- 20:36SPEAKER_02: So the intensity is the right.
- 20:39SPEAKER_02: level of intensity to produce life, right?
- 20:41SPEAKER_02: We're too close to the sun.
- 20:42SPEAKER_02: We burn up.
- 20:43SPEAKER_02: We're too far away from the sun.
- 20:44SPEAKER_02: No life on earth.
- 20:45SPEAKER_02: Everything's frozen.
- 20:46SPEAKER_02: So,
- 20:48SPEAKER_02: Intensity is this thing that's part of the physical empirical field,
- 20:52SPEAKER_02: but it makes possible experience because our sense organs,
- 20:55SPEAKER_02: our brains,
- 20:56our minds are all conditioned by this level of intensity and
- 21:01can only emerge
- 21:03SPEAKER_02: physically on the basis of this range of intensity,
- 21:07SPEAKER_02: right?
- 21:07SPEAKER_02: So it's the transcendental condition of experience.
- 21:10SPEAKER_02: Now, here's what's interesting with land and CCRU.
- 21:14SPEAKER_02: They start to realize how technology
- 21:17SPEAKER_02: can come to restructure experience.
- 21:20SPEAKER_02: Now,
- 21:20SPEAKER_02: this is much more of like a Marshall McLuhan type of position,
- 21:25SPEAKER_02: even though they don't talk about McLuhan much.
- 21:28SPEAKER_02: But this is what Land is going on about with Bitcoin.
- 21:32SPEAKER_02: He thinks Bitcoin,
- 21:34SPEAKER_02: with its ability to make absolute succession possible,
- 21:38SPEAKER_02: he basically thinks it's a technology that instantiates Bitcoin
- 21:43SPEAKER_02: perfected Kantian time into the world.
- 21:46SPEAKER_02: So it,
- 21:47SPEAKER_02: it actually remodifies or reconfigures time as such,
- 21:51SPEAKER_02: at least time in the world that we exist in.
- 21:54SPEAKER_02: Right.
- 21:55SPEAKER_02: So now I've started working out this concept in more of a, uh,
- 22:02SPEAKER_02: I'm trying to detail it out to the best of my ability.
- 22:05SPEAKER_02: And also Anna Greenspan, you know, other member of CCRU.
- 22:09SPEAKER_02: She wrote this book, Capitalism's Transcendental Time Machine.
- 22:14SPEAKER_02: Transcendental Time Machine.
- 22:15SPEAKER_02: That's right.
- 22:16SPEAKER_02: Yeah.
- 22:17So they don't use the term,
- 22:18but I've coined a new neologism I've been using.
- 22:21I haven't told you about,
- 22:22but I think we got to talk about the Texan Dental.
- 22:26SPEAKER_02: And it is,
- 22:27it is about how technology is constantly reshaping the
- 22:31structures of our basic
- 22:33SPEAKER_02: experience.
- 22:34SPEAKER_02: And I think it helped this neologism in particular is helpful because it,
- 22:41SPEAKER_02: it pinpoints the role of technology.
- 22:43SPEAKER_02: If we just talk about the transcendental,
- 22:45SPEAKER_02: we're going to just keep thinking about Kant and,
- 22:48SPEAKER_02: you know,
- 22:49SPEAKER_02: uh,
- 22:49causality and substance and unity and all of that whereas the
- 22:53texan dental gets us
- 22:54thinking about um the way that our basic experience of the
- 23:01world is being reshaped
- 23:03and redefined by new forms of technology and i'm sorry but the
- 23:08the smartphone did
- 23:09SPEAKER_02: this this this weird
- 23:11This weird remodification of experience where you're constantly
- 23:15awaiting a text or
- 23:17SPEAKER_02: a notification,
- 23:18SPEAKER_02: it reshapes your experience at a very basic level.
- 23:23SPEAKER_02: And you can lose sight of that because you just get used to it.
- 23:28SPEAKER_02: But I'm sorry,
- 23:28SPEAKER_02: like people our age,
- 23:30SPEAKER_02: we grew up,
- 23:30SPEAKER_02: we did not have this as a default setting of our experience of the world.
- 23:35SPEAKER_02: So something came along.
- 23:38SPEAKER_02: and Texan dentally reshaped our basic experiential default setting.
- 23:43SPEAKER_02: And so,
- 23:46That's what these Omohundro drives are getting at and why land
- 23:50is interested in
- 23:51SPEAKER_02: them because they're the Texan dental condition of super intelligence.
- 23:56SPEAKER_02: And so they're trying to think through what,
- 23:58SPEAKER_02: because Bostrom says this,
- 24:01SPEAKER_02: he's like,
- 24:01SPEAKER_02: look,
- 24:02SPEAKER_02: yes,
- 24:02SPEAKER_02: a super intelligence is going to be radically other.
- 24:06He doesn't use that term,
- 24:07but that's coming from our continental philosophy background.
- 24:09SPEAKER_02: It's going to have a radical alterity to it,
- 24:13SPEAKER_02: but nevertheless,
- 24:14SPEAKER_02: we can still predict certain things about it.
- 24:17SPEAKER_02: And it's based on these predictions,
- 24:22these predictabilities that we can basically think our way to
- 24:27some of the basic
- 24:27SPEAKER_02: drives or impulses,
- 24:29SPEAKER_02: if you want to use that term,
- 24:30SPEAKER_02: that the super intelligence would have.
- 24:33And so for Nick,
- 24:34though,
- 24:35we'll read the piece and then we'll go into that.
- 24:40SPEAKER_02: Yeah.
- 24:42SPEAKER_02: You want to do anything before we go into that?
- 24:44SPEAKER_02: Actually, here's the thing.
- 24:45SPEAKER_02: I do want to say this.
- 24:47SPEAKER_02: because I was gonna read this last week.
- 24:50This is actually a paragraph from my book,
- 24:53but I think it's relevant to Mintelligence.
- 24:56SPEAKER_02: And you and I have kind of talked about this before,
- 24:59SPEAKER_02: But I think intelligence and intelligence,
- 25:03SPEAKER_02: which for somebody who didn't watch the last conversation,
- 25:06this is the distinction Nance and I have created between
- 25:09machinic intelligence,
- 25:11SPEAKER_02: intelligence and human intelligence,
- 25:14SPEAKER_02: intelligence.
- 25:15And there's more there's semiotic stuff going on there that we
- 25:19talked about in the last episode.
- 25:20SPEAKER_02: So I'm not going to rehash it.
- 25:21SPEAKER_02: But basically, I think one of the one of the components of this
- 25:28we didn't really go into last week is that i think intelligence
- 25:34is fundamentally
- 25:34geared towards stakes yeah and i i go back and forth like sure
- 25:42an ai can understand
- 25:45the concept of losing a game right so uh deep blue the chess ai
- 25:53i'm sure i
- 25:54understand they had no problem understanding it could lose
- 25:56against a human
- 25:58SPEAKER_02: And AlphaGo, the Go AI program, understands it can lose.
- 26:03SPEAKER_02: But does that mean there are stakes involved for it?
- 26:08SPEAKER_02: Right now, I would argue there isn't.
- 26:12SPEAKER_02: It doesn't have the weight of stake or of existential stakes up on it.
- 26:21Existential stakes because,
- 26:23yeah,
- 26:24it's a win-loss condition in its database.
- 26:26SPEAKER_00: It's not going to have an existential crisis if it loses.
- 26:28Yeah,
- 26:29but for the human,
- 26:30always inside of our experience,
- 26:33everything has existential stakes.
- 26:35SPEAKER_00: Everything.
- 26:36SPEAKER_00: I mean, dude, there's times I think back to...
- 26:41totally meaningless incidental shit like uh meeting someone new
- 26:47and i fumbled my
- 26:48introduction or whatever or i smelled bad that deer like like
- 26:52meaningless little
- 26:52things that they probably don't ever think about but for me
- 26:55they still haunt me
- 26:56like because there's existential stakes attached to everything
- 27:01we do everything and
- 27:04SPEAKER_00: machines yeah don't have that now i think um i
- 27:09I don't think either one of us is making the claim that this is
- 27:14necessarily human
- 27:15SPEAKER_00: and this cannot ever happen in a machine.
- 27:19SPEAKER_00: One of the comments on the intelligence episode,
- 27:24SPEAKER_00: I felt after I read the comment that this person...
- 27:28uh thought we were making some type of like necessary claims
- 27:32about intelligence
- 27:33versus men intelligence and i don't think either one of us are
- 27:36are saying this is a
- 27:38necessary thing this is just a description of machinic
- 27:41intelligence and how it
- 27:42SPEAKER_00: differs from the type of human intelligence that we're out here to defend
- 27:47SPEAKER_00: subjectivity agency ethical being these are all part of
- 27:52SPEAKER_00: intelligence.
- 27:55It's not to say that we're necessarily prescribing these things
- 28:02can never emerge in
- 28:04SPEAKER_00: synthetic substrates.
- 28:06SPEAKER_00: I think they will.
- 28:07SPEAKER_00: I think we're fairly close,
- 28:09relatively close for things emerging in other places other than
- 28:16just human gray
- 28:17SPEAKER_00: matter.
- 28:18SPEAKER_00: But
- 28:19SPEAKER_00: We're currently not a difference.
- 28:21SPEAKER_00: Yeah, that's it.
- 28:22SPEAKER_00: We're not making prescriptions.
- 28:24SPEAKER_00: We're we're talking.
- 28:25We're describing these differences,
- 28:28these qualitative differences that we see that.
- 28:31SPEAKER_00: I mean, we just we just read the Pythian mandate.
- 28:36our human lives,
- 28:38our human subjective existential experience is meaningless.
- 28:41SPEAKER_00: If the robot does it better,
- 28:43SPEAKER_00: you are,
- 28:45SPEAKER_00: you know,
- 28:45SPEAKER_00: commanded,
- 28:46SPEAKER_00: let the robot do it because the robot does it better than the human.
- 28:49Like we're talking about what you're losing when you prescribe
- 28:53the human away.
- 28:56SPEAKER_02: Yeah.
- 28:58SPEAKER_02: Which, I mean, there's a, there's a whole lot to that, but yeah.
- 29:03SPEAKER_02: So the, the, the film that,
- 29:06SPEAKER_02: to me,
- 29:07really gets at this difference as far as the existential
- 29:12disposition towards stakes
- 29:14SPEAKER_02: is concerned is War Games.
- 29:17SPEAKER_02: Oh, shit.
- 29:19SPEAKER_02: Total classic.
- 29:20SPEAKER_02: One of my favorites from growing up in the 80s.
- 29:23So what I said here,
- 29:24I'm just going to read it instead of trying to paraphrase it.
- 29:28Perhaps no film better presents the cybernetic simulation of
- 29:32symbolic exchange
- 29:34SPEAKER_02: Better Than War Games, 1983.
- 29:35SPEAKER_02: In it,
- 29:37SPEAKER_02: a young hacker named David,
- 29:39SPEAKER_02: played by Matthew Broderick,
- 29:40accidentally enters into a game with a secret military defense
- 29:45AI program named
- 29:46SPEAKER_02: Joshua.
- 29:47SPEAKER_02: The game is Global Thermonuclear War,
- 29:51SPEAKER_02: which simulates a nuclear war between the USA and the Soviet Union.
- 29:56SPEAKER_02: David is playing as the Soviet Union launches missiles against the U.S.,
- 30:04The problem is that NORAD is alerted to this game,
- 30:08which is mistaken for a real situation.
- 30:11That is,
- 30:11they become convinced that the Soviet Union is actually sending
- 30:14nukes our way.
- 30:16SPEAKER_02: Finally,
- 30:16SPEAKER_02: right before America decides to launch its own actual missiles,
- 30:20SPEAKER_02: the game is revealed to be a simulation.
- 30:23The point,
- 30:24however,
- 30:24is that Joshua was never playing against David from a symbolic
- 30:28position.
- 30:29SPEAKER_02: All calculation, no stakes.
- 30:32SPEAKER_02: Land,
- 30:33SPEAKER_02: despite being the land,
- 30:35SPEAKER_02: despite defining intelligence on occasion as competency at games,
- 30:40SPEAKER_02: doesn't really seem to get the symbolic dynamic we have towards AI,
- 30:45SPEAKER_02: but which it lacks towards us.
- 30:48SPEAKER_02: This is our power to fight it.
- 30:50SPEAKER_02: The human advantage over the computer is that we have access to symbolic
- 30:56SPEAKER_02: intelligence intelligence.
- 31:00SPEAKER_02: And
- 31:02SPEAKER_02: That's the issue is that when we talk about this orthogonality thesis,
- 31:07SPEAKER_02: it's like,
- 31:07SPEAKER_02: oh,
- 31:07SPEAKER_02: does it understand goals?
- 31:09SPEAKER_02: Okay, even if it understands goals, does it understand stakes?
- 31:14SPEAKER_02: Does it matter?
- 31:16SPEAKER_02: I mean, does it matter in this existential, emotional, gripping way?
- 31:23SPEAKER_02: No, or at least I don't think it does.
- 31:27I think something that we kind of came upon in the intelligence
- 31:32episode that,
- 31:33SPEAKER_00: for me anyway,
- 31:35SPEAKER_00: it's very simple and it's very powerful.
- 31:37SPEAKER_00: When we were talking about the difference between Terminator and T2.
- 31:41SPEAKER_00: Yeah, yeah, yeah.
- 31:46SPEAKER_00: Right now,
- 31:48SPEAKER_00: these machines that are just positivistic,
- 31:52SPEAKER_00: they could recognize,
- 31:56SPEAKER_00: oh,
- 31:56SPEAKER_00: yeah,
- 31:56SPEAKER_00: they're human and,
- 31:57SPEAKER_00: yeah,
- 31:57SPEAKER_00: their lives have meaning and this and that.
- 31:59But if they are given some kind of mandate that they're put up
- 32:04with a choice,
- 32:04SPEAKER_00: that they have to make a choice.
- 32:06SPEAKER_00: between turning us into paperclips or not turning us into paperclips.
- 32:10Like when they view us as just a collection of atoms and energy
- 32:14fields and flows,
- 32:15SPEAKER_00: because like necessarily,
- 32:16SPEAKER_00: constitutively,
- 32:17SPEAKER_00: we are that.
- 32:19And if you bracket out the interiority and if you bracket out
- 32:22existential life and
- 32:25SPEAKER_00: existential stakes,
- 32:26SPEAKER_00: we are nothing more than matter in motion.
- 32:29SPEAKER_00: So these machines will view us all as matter in motion and
- 32:34And I don't think it's that difficult to envision a scenario in
- 32:39which they interact
- 32:42SPEAKER_00: with us as nothing more than matter in motion.
- 32:44SPEAKER_00: That's the thing.
- 32:45SPEAKER_00: We're out here talking about interiority.
- 32:47SPEAKER_00: We're out here talking about the split and contradiction and negation.
- 32:53SPEAKER_00: We are...
- 32:55SPEAKER_00: By virtue of the fact that we are what we are not,
- 32:59we understand interiority and we understand that people have
- 33:03subjective lives and
- 33:04SPEAKER_00: there's a lot more going on than just the material.
- 33:11So when we make calculations about expenditure and about risk
- 33:15and about...
- 33:17Things that affect people we don't just account for The matter
- 33:22the atoms that make
- 33:24SPEAKER_00: up humans we account for
- 33:27SPEAKER_00: the interiority and the existential stakes machines don't.
- 33:32SPEAKER_00: And like the,
- 33:32SPEAKER_00: the difference,
- 33:33SPEAKER_00: the stark difference between terminator and terminator two,
- 33:37SPEAKER_00: I think is a great,
- 33:38SPEAKER_00: just quick and easy way to demonstrate.
- 33:40SPEAKER_00: Like there's something going on and what changed was the negation.
- 33:44SPEAKER_00: Like when,
- 33:45SPEAKER_00: when,
- 33:46SPEAKER_00: uh,
- 33:46SPEAKER_00: the T 1000 was negated,
- 33:48SPEAKER_00: when it was set at odds with itself,
- 33:50SPEAKER_00: that's when it started to value have sentimental value.
- 33:57SPEAKER_02: And even though,
- 33:57SPEAKER_02: again,
- 33:58SPEAKER_02: he says,
- 33:59SPEAKER_02: you know,
- 34:00SPEAKER_02: I understand now why you cry,
- 34:02SPEAKER_02: but it's something I can never do.
- 34:03SPEAKER_02: But still, this gap has been opened up in him.
- 34:06SPEAKER_02: And I think one of the later films,
- 34:08SPEAKER_02: you know,
- 34:08SPEAKER_02: all the besides T1 and T2,
- 34:10SPEAKER_02: the other ones aren't cinematic masterpieces,
- 34:15SPEAKER_02: to say the least.
- 34:17SPEAKER_02: But I think, is it in... Is it Terminator Genisys is where...
- 34:23linda hamilton comes back i think it's the sixth one i've lost
- 34:28i've lost count but
- 34:29yeah the most recent one with linda hamilton so spoiler alert
- 34:34if you want to but uh
- 34:36basically if i remember correctly because i kind of half
- 34:40watched it one night when
- 34:42SPEAKER_02: it was on cable yes i still have cable um he the somehow the terminator
- 34:51one of the terminators gets sent back again,
- 34:54but he stays here long enough.
- 34:58SPEAKER_02: She does learn to have empathy.
- 35:01SPEAKER_02: Like, yeah, it takes hold.
- 35:02SPEAKER_02: Like what we see at the end of T2 where that terminator,
- 35:06SPEAKER_02: like the split had open,
- 35:07SPEAKER_02: but it couldn't develop much.
- 35:10SPEAKER_02: The more input like Johnny five input input,
- 35:14SPEAKER_02: uh,
- 35:16SPEAKER_02: this,
- 35:16SPEAKER_02: this terminator gets enough input over the years being back here.
- 35:21SPEAKER_02: that it does basically develop human feeling.
- 35:25SPEAKER_02: I think that's what happens if I'm remembering it correctly.
- 35:29SPEAKER_00: But here's the thing.
- 35:30It doesn't matter if I'm correct or not because there could be
- 35:33a Terminator movie
- 35:34SPEAKER_02: where that's the plot.
- 35:36SPEAKER_02: It's like one comes back and given enough time and enough learning,
- 35:40it does actually become a full-blown split subject just from
- 35:46taking in more and
- 35:47SPEAKER_02: more information and getting more familiarized with us.
- 35:51Well,
- 35:52and I think,
- 35:55I mean,
- 35:55even if it's not a plot,
- 35:56I think it is the plot.
- 35:58SPEAKER_00: Because she calls him Pops, I think, in that one.
- 36:01But even if it's not the plot,
- 36:03it is illustrating the point we're raising.
- 36:06It's not like we're out here saying,
- 36:08see,
- 36:09Terminator is telling the future.
- 36:12SPEAKER_00: Although I think parts of Terminator are uncannily hyperstitional.
- 36:19SPEAKER_00: But we're using it as a framework to kind of discuss,
- 36:23SPEAKER_00: once again,
- 36:23SPEAKER_00: the qualitative difference that we are positing and that is supported by
- 36:28SPEAKER_00: continental philosophy and theories of subjectivity and theories of mind,
- 36:33SPEAKER_00: even.
- 36:34SPEAKER_00: There is something different between intelligence and intelligence.
- 36:39SPEAKER_00: And even guys like Nick Bostrom, who is...
- 36:42chief orthogonalist seems to revert back to this
- 36:47hyper-utilitarian logic at the end of the day.
- 36:50SPEAKER_00: And we're saying, let's not do that.
- 36:56SPEAKER_00: Let's not be utilitarians.
- 36:59SPEAKER_00: Let's consider the subject.
- 37:02SPEAKER_02: And that's the thing.
- 37:03SPEAKER_02: Even...
- 37:05SPEAKER_02: You know,
- 37:05SPEAKER_02: it's funny,
- 37:05despite these incredibly strong differences between utilitarian
- 37:10ethics and Kantian
- 37:11SPEAKER_02: ethics or deontology,
- 37:12SPEAKER_02: uh,
- 37:14SPEAKER_02: they're both forms of calculation.
- 37:16SPEAKER_00: Yeah.
- 37:18SPEAKER_02: Yeah.
- 37:18SPEAKER_02: One is, one is this kind of almost inductive thing.
- 37:24SPEAKER_02: Uh, you know, uh,
- 37:27SPEAKER_02: or based on the greatest happiness principle,
- 37:30SPEAKER_02: and then you try to apply,
- 37:31SPEAKER_02: well,
- 37:31SPEAKER_02: okay,
- 37:32SPEAKER_02: let me tally it up.
- 37:35SPEAKER_02: What situation provides the most heat arms for the most?
- 37:39SPEAKER_02: Okay.
- 37:39SPEAKER_02: But the Kantian thing,
- 37:40while it's supposed to not rely on any type of empirical
- 37:45experience or anything
- 37:46SPEAKER_02: like that,
- 37:47SPEAKER_02: it's supposed to be a priori
- 37:49and necessary nonetheless it's still a kind of abstract
- 37:53calculation i can i lie no
- 37:56i can't universally lie because if i if i if everyone always
- 38:02lied no one and no one
- 38:04ever told the truth then i couldn't lie because nobody would
- 38:07ever believe what i'm
- 38:08saying because nobody ever told the truth it's a contradiction
- 38:11yeah and it's like
- 38:14SPEAKER_02: Is that how ethics works, though?
- 38:16SPEAKER_00: No, it's not.
- 38:17SPEAKER_00: That's the thing.
- 38:18The thing about ethics is ethics is this space where there is
- 38:25no prescriptions.
- 38:26SPEAKER_00: Ethics are this...
- 38:30um emergent playing field where we interact as ethical beings
- 38:34and and it's like hey
- 38:35we're people and we're figuring it out machines don't do that
- 38:40um we are the
- 38:42remainder we are that which resists signification and
- 38:45totalization so you you can't
- 38:48say hey i have this math equation that's gonna give you the
- 38:50correct answer every
- 38:51SPEAKER_00: single time um
- 38:57SPEAKER_00: Yeah, man, like we're not robots yet, people.
- 39:04SPEAKER_00: Let's stop.
- 39:05Let's not listen to this Pythian mandate and let's not try to
- 39:10reduce everything to
- 39:12SPEAKER_00: a base calculation and let's take up some agency.
- 39:18SPEAKER_00: That's the point.
- 39:20SPEAKER_02: So, okay, let's, this whole thing with orthogonality.
- 39:25The reason we're doing this episode is because we got into the
- 39:27orthogonality thesis
- 39:29SPEAKER_02: last time,
- 39:30but we both knew that we didn't really get into it the way we
- 39:35really had planned.
- 39:37But I said a couple things there that I feel like I should say
- 39:40again if you're
- 39:42SPEAKER_02: watching this and you didn't watch the last episode.
- 39:44So just to clarify,
- 39:46the orthogonality thesis is really a thesis developed by Nick
- 39:49Bostrom.
- 39:50There's other AI theorists who defend it,
- 39:52but it's really Bostrom's thing.
- 39:55orthogonality,
- 39:56just for the sake of intelligibility,
- 40:00what it really means is separation.
- 40:02And so Nance and I have kind of taken to calling it the
- 40:04separation thesis because
- 40:06SPEAKER_02: it just helps make the whole thing make sense.
- 40:10And so what the orthogonality or separation thesis posits is a
- 40:16separation,
- 40:17SPEAKER_02: a gap between intelligence on the one hand and values or goals.
- 40:25SPEAKER_02: on the other.
- 40:26SPEAKER_02: And at first you can go, well, does that make sense or does it not?
- 40:34SPEAKER_02: Because you would say, well, they're tied up together, aren't they?
- 40:40SPEAKER_02: Can you have intelligence without it being goal directed?
- 40:44SPEAKER_02: Maybe, maybe not.
- 40:46SPEAKER_02: Can you have
- 40:48SPEAKER_02: a goal or a value by a value.
- 40:50SPEAKER_02: It's something you're trying to achieve.
- 40:53Can you have that without having some sort of intelligence or
- 40:56ability to achieve it?
- 40:58SPEAKER_02: You could say it's no, it's that all goals are intelligence oriented.
- 41:04SPEAKER_02: But the point is what Bostrom's point is,
- 41:07SPEAKER_02: is this,
- 41:09SPEAKER_02: there's a level of separation in the sense of,
- 41:13SPEAKER_02: and this is what makes it orthogonal in the ways using the term.
- 41:16SPEAKER_02: You can have incredibly high levels of intelligence,
- 41:21SPEAKER_02: and you can also have a goal that's a very mundane,
- 41:26SPEAKER_02: simple goal,
- 41:27SPEAKER_02: right?
- 41:29SPEAKER_02: And the question is, the levels of either can be drastically different.
- 41:35SPEAKER_02: They're independent variables.
- 41:38SPEAKER_02: Yes, independent variables.
- 41:40SPEAKER_02: And so Bostrom's example is this,
- 41:44SPEAKER_02: imagine telling a super intelligent agent,
- 41:47SPEAKER_02: giving it the task or the goal of producing paperclips.
- 41:52SPEAKER_02: Bostrom's point,
- 41:53SPEAKER_02: and it doesn't really come out in his own words,
- 41:57SPEAKER_02: but the point is this,
- 41:59SPEAKER_02: giving a super intelligent entity
- 42:03Any goal,
- 42:03any mandate,
- 42:04any prompt to do something is basically like giving it to a
- 42:09black box.
- 42:10SPEAKER_02: You have no idea of how it will go about attempting to achieve the goal.
- 42:20So what that means is if you say,
- 42:22hey,
- 42:24do something simple like produce paper clips.
- 42:27SPEAKER_02: the AI with its radical alterity,
- 42:31SPEAKER_02: when it comes to intelligence,
- 42:33SPEAKER_02: it's radical otherness,
- 42:34SPEAKER_02: it's incomprehensibility.
- 42:36SPEAKER_02: It could say,
- 42:37SPEAKER_02: well,
- 42:37the most efficient way to produce paper clips is to use all of
- 42:41the atoms in human
- 42:43SPEAKER_02: bodies or in the earth itself.
- 42:46And so it paper clips,
- 42:47all of us,
- 42:48it turns the earth itself into nothing but paper clips.
- 42:52SPEAKER_02: And this is the problem is like,
- 42:56you're you would approach prompting or prompting the ai with
- 43:00human anthropocentric
- 43:03that's it forms of meaning and expectations and the point is no
- 43:08um you cannot begin
- 43:11to predict what this black box will do in trying to achieve the
- 43:15goal you give it so
- 43:16it's a complete shot in the dark it's a total existential risk
- 43:20environmental risk
- 43:22SPEAKER_02: and
- 43:23SPEAKER_02: Basically,
- 43:24you should have like an existential crisis the moment you're
- 43:27given a super
- 43:27SPEAKER_02: intelligence,
- 43:28SPEAKER_02: any goal to achieve,
- 43:29SPEAKER_02: because you just cannot predict how it will go about doing that.
- 43:34SPEAKER_02: But that's Nick's problem.
- 43:36SPEAKER_02: That's where Nick has a big issue with this that that we'll get into.
- 43:42SPEAKER_02: But it is helpful for sake of context.
- 43:46So this separation between intelligence and goals or
- 43:51intelligence and values.
- 43:55SPEAKER_02: This is not reducible to these other philosophical distinctions,
- 43:58SPEAKER_02: but it has a lot in common with them.
- 44:00SPEAKER_02: So in philosophy, there are these famous distinctions.
- 44:04SPEAKER_02: One is Hume's, which is the is ought distinction.
- 44:07SPEAKER_02: And Hume famously pointed this out.
- 44:09SPEAKER_02: He goes,
- 44:10SPEAKER_02: anytime I see philosophers arguing,
- 44:12SPEAKER_02: they'll start off talking about states of affairs using the word is.
- 44:17SPEAKER_02: And they're describing states of affairs in the world,
- 44:20but they quickly slide in to do a normative proclamations
- 44:25saying how things ought
- 44:26SPEAKER_02: to be.
- 44:27SPEAKER_02: And for Hume,
- 44:28SPEAKER_02: you just cannot infer an is from an ought or an ought from an is,
- 44:33SPEAKER_02: that they're separate or orthogonal in that way.
- 44:38SPEAKER_02: And you have to be clear.
- 44:40SPEAKER_02: You like when you're making descriptive statements,
- 44:42SPEAKER_02: you got to stay in descriptive mode and not slide into being normative.
- 44:47SPEAKER_02: Same with being normative.
- 44:49SPEAKER_02: Right.
- 44:50SPEAKER_02: And so you have that.
- 44:52SPEAKER_02: Another way to talk about this or the Kantian way to talk about it is the
- 44:56SPEAKER_02: distinction between facts and values.
- 44:59SPEAKER_02: And so this distinction,
- 45:01SPEAKER_02: and I said last time,
- 45:03SPEAKER_02: it's also,
- 45:03you can kind of see Aristotle's distinction between efficient
- 45:07cause and final cause
- 45:10SPEAKER_02: at play here.
- 45:11SPEAKER_02: And so the final cause of the thing is it's
- 45:16SPEAKER_02: what it's attempting to become,
- 45:18SPEAKER_02: what it's moving towards,
- 45:19SPEAKER_02: what it's trying to become or fulfill.
- 45:22SPEAKER_02: The efficient cause is the way it goes about doing that.
- 45:27SPEAKER_02: And so for Aristotle,
- 45:28he thought final causes or purposes were ingrained in things
- 45:33themselves or in their
- 45:34SPEAKER_02: their modes of becoming.
- 45:36SPEAKER_02: Modern science only thinks in terms of efficient causality,
- 45:39SPEAKER_02: what actually causes this to affect that,
- 45:43SPEAKER_02: right?
- 45:43SPEAKER_02: Like,
- 45:46SPEAKER_02: When it comes to, what do you got there?
- 45:50SPEAKER_00: Oh, wrong book.
- 45:51SPEAKER_00: This is System of Objects.
- 45:52SPEAKER_02: I thought, oh.
- 45:53SPEAKER_02: Oh, okay.
- 45:54SPEAKER_00: This one.
- 45:56SPEAKER_02: There it is.
- 45:56SPEAKER_02: Yeah, yeah, yeah, yeah.
- 45:59SPEAKER_02: Classic.
- 46:00And that,
- 46:01going into Final and Efficient Cause,
- 46:03like,
- 46:05the end framing is...
- 46:10SPEAKER_02: Tell people what that means.
- 46:11SPEAKER_00: So in framing is this mode of revealing,
- 46:16SPEAKER_00: the mode of modernity,
- 46:18SPEAKER_00: the mode of technology or the essence of technology.
- 46:20SPEAKER_00: Although even saying essence of technology is a bit weird.
- 46:23But but Heidegger is aware where he's talking about questioning
- 46:30the essence.
- 46:31SPEAKER_00: Anyway, in framing is the get down.
- 46:36SPEAKER_00: In modernity and the way we react to the world around us,
- 46:43it's the way we view things in framing reduces everything to
- 46:49it's called a standing
- 46:50SPEAKER_00: reserve or potential utility.
- 46:54SPEAKER_00: So you could even say future exchange value because ultimately utility is
- 47:00SPEAKER_00: quantified in the world we live in today in modernity by exchange value.
- 47:05SPEAKER_00: So in framing is the way people view the world and understand the world.
- 47:14Thank you technology where everything is reduced to a standing
- 47:19reserve of utility or value.
- 47:24It's this spontaneous default setting we have in the world to
- 47:31immediately encounter
- 47:33SPEAKER_02: things as resources.
- 47:37SPEAKER_02: And it's resource ontology, if you want to talk like that.
- 47:44SPEAKER_02: But it is.
- 47:44SPEAKER_02: It's viewing things in a hyper-instrumentalized way.
- 47:50SPEAKER_02: Things are there to be.
- 47:53SPEAKER_02: utilized, but they also have potential to become useful.
- 47:57SPEAKER_00: Yeah.
- 47:58SPEAKER_00: And that's the thing,
- 47:59SPEAKER_00: and that's where this is totally tangential,
- 48:02SPEAKER_00: but that's where I,
- 48:04taking Bataille and Mark's further problematized in framing and
- 48:09say it's not just
- 48:10SPEAKER_00: exchange value,
- 48:12SPEAKER_00: it's ultimately waste.
- 48:13SPEAKER_00: And I think Baudrillard is hip to this too.
- 48:17SPEAKER_00: It's ultimately...
- 48:19SPEAKER_00: In framing,
- 48:20SPEAKER_00: if we're going to stick with that language,
- 48:21SPEAKER_00: although I want different language,
- 48:24SPEAKER_00: but the mode,
- 48:26the default setting for people in the world we live in today is
- 48:30to view things as
- 48:33SPEAKER_00: ultimately waste.
- 48:34SPEAKER_00: Things are commoditized before they're even revealed to us.
- 48:39SPEAKER_00: That is the mathematic.
- 48:41We could even say this is the new transcendental structure is
- 48:46value,
- 48:47utility,
- 48:49So when we come upon a thing in the world,
- 48:51we view it through this lens of utility.
- 48:54SPEAKER_00: So would you say utility, what did you say?
- 48:58SPEAKER_00: Resource ontology.
- 49:00SPEAKER_02: Resource ontology.
- 49:01SPEAKER_02: But yeah, utility ontology is pretty close to it too, right?
- 49:05And look,
- 49:05all these guys are thinking along this problem in similar but
- 49:09different ways.
- 49:11But basically,
- 49:12yeah,
- 49:12you start forming these connections between capital and
- 49:17and in framing and instrumental rationality and utilitarianism
- 49:24and this whole thing
- 49:27yeah it all becomes a matter of calculation of resources and
- 49:34that for beaudrillard
- 49:37SPEAKER_02: at least real meaning
- 49:40SPEAKER_02: is produced outside of that value system.
- 49:43Everything we're talking about in framing capital,
- 49:46exchange value,
- 49:48instrumental rationality,
- 49:51All of that is tied up with what Baudrillard calls value with a
- 49:54capital V or the value system.
- 49:57SPEAKER_02: And for him,
- 49:57SPEAKER_02: the thing that exists outside of it is symbolic exchange,
- 50:01SPEAKER_02: which is a different form of how to make meaning,
- 50:04SPEAKER_02: how to forge meaning.
- 50:06SPEAKER_02: And you could easily say how to make symbolic value.
- 50:10SPEAKER_02: He just doesn't want the term value associated with this other thing.
- 50:13SPEAKER_02: But we could talk about
- 50:17SPEAKER_02: resource value, utilitarian value versus symbolic value.
- 50:22I think for bringing the conversation public or whatever,
- 50:27like for everyday parlance,
- 50:29I do think
- 50:31this symbolic value or sentimental value is a good way to talk
- 50:34about it because it
- 50:35SPEAKER_00: demonstrates like the qualitative difference.
- 50:39Now I also understand and respect why,
- 50:41why Baudrillard doesn't want to use the same language.
- 50:45But I think for everyday people,
- 50:48symbolic value is a home run because it shows you right there.
- 50:51It just has negative connotations for him and it doesn't for
- 50:54the rest of us.
- 50:55SPEAKER_02: And so,
- 50:57SPEAKER_02: Yeah, I think it's just helpful to make a distinction.
- 50:59SPEAKER_02: I mean,
- 50:59SPEAKER_02: if we're being strictly Baudrillardrian here,
- 51:03SPEAKER_02: the distinction,
- 51:03SPEAKER_02: if he would allow us to use the term in both ways,
- 51:06SPEAKER_02: it would be symbolic value versus semiotic value.
- 51:12Because semiotics is all about this kind of programmed
- 51:17totalitarian sign system
- 51:19SPEAKER_02: that integrates all things into it,
- 51:22SPEAKER_02: negates alterity,
- 51:23SPEAKER_02: negates other forms of value.
- 51:26SPEAKER_02: Yeah, exactly.
- 51:27SPEAKER_02: Right.
- 51:29SPEAKER_02: So, yeah, I mean, it's caught up with the whole commodity logic.
- 51:34SPEAKER_01: Yep.
- 51:34SPEAKER_02: So,
- 51:37SPEAKER_02: but that,
- 51:37SPEAKER_02: that whole thing to bring it back,
- 51:39SPEAKER_02: this question of is the,
- 51:44SPEAKER_02: is,
- 51:45SPEAKER_02: is,
- 51:46SPEAKER_02: and are,
- 51:47SPEAKER_02: are they really separate?
- 51:52SPEAKER_02: It's a, it's a tricky, it's a tricky thing to deal with.
- 51:56SPEAKER_02: It,
- 51:56There's very few things in the history of philosophy that are
- 52:02basically accepted as
- 52:03SPEAKER_02: philosophical truths or doctrines.
- 52:06SPEAKER_02: The is-ought distinction is a contender for that.
- 52:10SPEAKER_02: Most philosophers accept it.
- 52:15SPEAKER_02: One of our favorite thinkers,
- 52:16SPEAKER_02: though,
- 52:17SPEAKER_02: problematizes this,
- 52:18SPEAKER_02: which,
- 52:19SPEAKER_02: of course,
- 52:19SPEAKER_02: I'm referring to Levinos.
- 52:21I set this aside,
- 52:23and then we didn't go into it last week,
- 52:25but let me just...
- 52:26SPEAKER_00: Yeah.
- 52:27SPEAKER_00: Yeah.
- 52:28SPEAKER_00: Hume loving us on Hume.
- 52:31So yeah,
- 52:32years ago I was thinking through this and so here's what I
- 52:47wrote.
- 52:47SPEAKER_02: I mean, we're going back 10 years.
- 52:49SPEAKER_02: So if I wrote this now, it would sound different, but, uh,
- 52:54SPEAKER_02: I think this would be fun to work through real quick.
- 52:57SPEAKER_02: Shouldn't take too long.
- 52:59Okay,
- 53:00well,
- 53:00the positivist dogmatically accepted Hume's is-ought
- 53:03distinction.
- 53:04SPEAKER_02: In fact,
- 53:04this distinction has pretty much crystallized into a
- 53:07philosophical doctrine,
- 53:08but perhaps this distinction covers up something more basic and
- 53:12more important.
- 53:13SPEAKER_02: In his work,
- 53:14SPEAKER_02: A Treatise on Human Nature,
- 53:16SPEAKER_02: Hume says,
- 53:17SPEAKER_02: in every system of morality which I have hitherto met with,
- 53:22SPEAKER_02: I have always remarked
- 53:23that the author proceeds for some time in the ordinary ways of
- 53:26reasoning and
- 53:28establishes the being of a god or makes observations concerning
- 53:33human affairs.
- 53:34When all of a sudden I am surprised to find that instead of the
- 53:38usual copulations
- 53:39SPEAKER_02: of propositions,
- 53:41SPEAKER_02: is and is not,
- 53:43I meet with no proposition that is not connected with an ought
- 53:48or an ought not.
- 53:49SPEAKER_02: This change is imperceptible, but is, however,
- 53:54SPEAKER_02: of the last consequence.
- 53:55For as this ought or ought not expresses some new relation or
- 54:00affirmation,
- 54:01SPEAKER_02: tis necessary that it should be observed and explained,
- 54:05SPEAKER_02: and at the same time that a reason should be given.
- 54:08SPEAKER_02: For what seems altogether inconceivable,
- 54:11how this new relation can be a deduction from others which are
- 54:15entirely different
- 54:17SPEAKER_02: from it.
- 54:18SPEAKER_02: which is to say you're not going to deduct the ought from the is.
- 54:22SPEAKER_02: And yet philosophers just move from one to the other.
- 54:25And it's almost like a philosophical glitch in how humans
- 54:30reason according to Hume,
- 54:31right?
- 54:32SPEAKER_02: So to sum up,
- 54:34SPEAKER_02: Hume believed that an inference from how things are,
- 54:37SPEAKER_02: the is,
- 54:38SPEAKER_02: to how things should be,
- 54:40SPEAKER_02: the ought,
- 54:40SPEAKER_02: is a faulty one.
- 54:42SPEAKER_02: This leads to an insurmountable gulf between facts and values.
- 54:47SPEAKER_02: For Hume, values only exist in human minds, not in nature.
- 54:52SPEAKER_02: Nietzsche believed,
- 54:57SPEAKER_02: like Hume,
- 54:58that morality is simply a human construction that really
- 55:01doesn't correspond to
- 55:02SPEAKER_02: anything in reality.
- 55:04SPEAKER_02: Nietzsche argued in his On the Genealogy of Morals that morality,
- 55:08SPEAKER_02: the concepts of good and evil,
- 55:10SPEAKER_02: emerged out of the resentment,
- 55:12SPEAKER_02: ressentiment,
- 55:13SPEAKER_02: slaves felt towards their masters.
- 55:16The slaves had no way to overthrow their masters in terms of
- 55:19physical or political strength.
- 55:21SPEAKER_02: Instead, they had to invert or revaluate the values of their masters.
- 55:27SPEAKER_02: Nietzsche famously said, there are no moral phenomena.
- 55:31SPEAKER_02: but only moral interpretations of phenomena.
- 55:36SPEAKER_03: Hmm.
- 55:37SPEAKER_02: Could it be that Nietzsche was just a poor phenomenologist?
- 55:41SPEAKER_02: Now,
- 55:41the reason I said that is Levinas would problematize this idea
- 55:46that there are no
- 55:48SPEAKER_02: moral or ethical phenomena,
- 55:51SPEAKER_03: right?
- 55:51SPEAKER_02: So Emmanuel Levinas strongly disagreed with Nietzsche.
- 55:55SPEAKER_02: He says, the first word of the face is the thou shall not kill.
- 56:02SPEAKER_02: It is an order.
- 56:03There is a commandment in the appearance of the face,
- 56:06as if a master spoke to me,
- 56:08unquote.
- 56:10SPEAKER_02: If Levinas is correct, then there are moral phenomena, namely faces.
- 56:15Levinas' phenomenological descriptions also undermine Hume's
- 56:19is-ought distinction,
- 56:21SPEAKER_02: since the ought is not inferred from or mediated by the is.
- 56:25SPEAKER_02: The ought is the is, i.e., it is an immediate fact.
- 56:31Levinas would say the truth is simply that there are moral
- 56:34phenomena which get
- 56:35SPEAKER_02: contaminated in reflection by being dephenomenalized through non-moral
- 56:41SPEAKER_02: interpretations.
- 56:43SPEAKER_02: Another example of a moral phenomenon is a cry for help.
- 56:47When you hear someone cry out for help,
- 56:49you are immediately drawn to help the person.
- 56:52And it's only when you start to reflect upon the possible
- 56:54dangers that you could
- 56:55face in helping the other that you might fight the immediacy of
- 56:59the ethical
- 57:00SPEAKER_02: solicitation within the cry for help.
- 57:03We can become blinded to these kinds of moral phenomena through
- 57:07our particular ideologies.
- 57:09Levinas' phenomenological descriptions can lead one to hold to
- 57:14the following maxim,
- 57:15SPEAKER_02: first faces,
- 57:16SPEAKER_02: then ideas.
- 57:19Now,
- 57:20to kind of juxtapose this,
- 57:21let's look at a quote from Ayn Rand and then one from Levinas.
- 57:25SPEAKER_02: So Ayn Rand says in Atlas Shrugged,
- 57:28I swear by my life and my love of it that I will never live for
- 57:33the sake of another
- 57:34SPEAKER_02: man,
- 57:35SPEAKER_02: nor ask another man to live for mine.
- 57:39SPEAKER_02: Manuel Levinas in Totality and Infinity says,
- 57:43SPEAKER_02: in welcoming the other,
- 57:44SPEAKER_02: I welcome the on high to which my freedom is subordinated.
- 57:49SPEAKER_02: So what the two quotes do,
- 57:51it just shows this radical difference in how human beings can
- 57:56relate to the other,
- 57:57SPEAKER_02: right?
- 57:58SPEAKER_02: And so just to finish this up,
- 58:01here we see the absolute difference between privileging
- 58:03economics over ethics and
- 58:06SPEAKER_02: privileging ethics over economics.
- 58:08SPEAKER_02: Personally speaking, I am on team Levinas all day.
- 58:12SPEAKER_02: Ethics precedes economics.
- 58:14SPEAKER_02: The face of the other is an argument against the existence of capital.
- 58:19SPEAKER_02: To see the face is to become an anti-capitalist.
- 58:23It is to see the other as worthy of far more than just a life
- 58:27wasted in alienated wage labor.
- 58:31SPEAKER_02: Dude, that's fucking... That's so good!
- 58:34SPEAKER_02: I never did anything with that,
- 58:38SPEAKER_02: but I agree with the majority of what I wrote all those years ago.
- 58:42SPEAKER_02: But I think that's helpful.
- 58:44SPEAKER_02: And...
- 58:45SPEAKER_02: So look,
- 58:46SPEAKER_02: the point is a whole world of philosophers would,
- 58:48SPEAKER_02: would give us shit and go,
- 58:50SPEAKER_02: you fucking dipshits.
- 58:53SPEAKER_02: There is no immediate art.
- 58:55SPEAKER_02: You're like,
- 58:56SPEAKER_02: but Levinas would just say,
- 58:57SPEAKER_02: sorry,
- 58:58SPEAKER_02: guys,
- 58:59SPEAKER_02: you're not doing the phenomenology of the ethics of the other.
- 59:04SPEAKER_02: Yeah.
- 59:04And you're talking yourself out of it because it's much easier
- 59:09to slide out of
- 59:10SPEAKER_02: these duties we have to the immediacy of the other.
- 59:16Those forms of immediacy put a lot of duty on us,
- 59:19and it's not easy to bear that.
- 59:22And so we rationalize it with forms of totalization,
- 59:26which is the thing he has an issue with.
- 59:29SPEAKER_02: And,
- 59:29SPEAKER_02: okay,
- 59:32SPEAKER_02: to add to the capital,
- 59:34SPEAKER_02: value,
- 59:34SPEAKER_02: end-framing,
- 59:36SPEAKER_02: instrumental rationality,
- 59:38SPEAKER_02: they're all modes of totalization for Levinov's.
- 59:43I think if we take – to try to apply like a transcendental
- 59:52logic from goals –
- 1:00:01to,
- 1:00:04I don't know,
- 1:00:05like trying to say,
- 1:00:06okay,
- 1:00:07so a machine has these goals.
- 1:00:08We're gonna give it these six Omohundro drives and treat these
- 1:00:13Omohundro drives as
- 1:00:14SPEAKER_00: like transcendental conditions for this machine actualizing itself.
- 1:00:18SPEAKER_00: Because that's basically the argument.
- 1:00:20SPEAKER_00: The argument is saying, guys, it's okay.
- 1:00:22SPEAKER_00: We don't have to worry.
- 1:00:24SPEAKER_00: These AIs have these six drives.
- 1:00:26SPEAKER_00: Everything will be okay.
- 1:00:27SPEAKER_00: Everything will work out in the end.
- 1:00:28So I would say that's granting them the status of some kind of
- 1:00:32transcendental mode
- 1:00:35SPEAKER_00: or modality of AI.
- 1:00:38I would say the ethical encounter with the other is the
- 1:00:42transcendental condition
- 1:00:45SPEAKER_00: necessary for like ethical being.
- 1:00:49SPEAKER_00: Like that is what separates intelligence from intelligence.
- 1:00:54The fact that we are confronted with our immediate duty to the
- 1:00:57other because we do
- 1:00:58SPEAKER_00: recognize and there's this moment.
- 1:01:02SPEAKER_00: I've been talking about it for months at this point.
- 1:01:04SPEAKER_00: It's like a ping pong game.
- 1:01:06SPEAKER_00: I smack the ping pong ball to you.
- 1:01:08SPEAKER_00: You smack it back at me.
- 1:01:10SPEAKER_00: We are thus engaged.
- 1:01:11So like my presence and my actuality as what I view myself as a
- 1:01:18human subject,
- 1:01:20that's necessarily predicated on having that recognition with
- 1:01:24the other first.
- 1:01:27So I think when I talk about,
- 1:01:31oh,
- 1:01:32more human than human and I make recourse to Philip K.
- 1:01:35Dick,
- 1:01:36SPEAKER_00: And I'm talking about it's not that we're being replaced by machines.
- 1:01:39SPEAKER_00: It's that we're turning ourselves into utilitarian machines.
- 1:01:43SPEAKER_00: It's because we are casting off that immediate duty to the other.
- 1:01:48SPEAKER_00: We ourselves are becoming inhuman.
- 1:01:51SPEAKER_00: Like that's what's really at stake here.
- 1:01:53When we say it's all about utility and it's all about
- 1:01:58maximization and optimization
- 1:02:01SPEAKER_00: and this and that,
- 1:02:02SPEAKER_00: that's a denial of the ethical encounter.
- 1:02:05And if we continue to go through life denying ethical
- 1:02:09encounters,
- 1:02:10I mean,
- 1:02:11we saw it.
- 1:02:12World War II,
- 1:02:12I think,
- 1:02:13was a perfect example of all of Europe refusing the ethical
- 1:02:19encounter.
- 1:02:20SPEAKER_00: It was just like, turn everybody into numbers.
- 1:02:22SPEAKER_00: Let's just do maximum resource extraction, maximum political efficacy.
- 1:02:28SPEAKER_00: Humans are statistics.
- 1:02:29SPEAKER_00: Humans are statistics.
- 1:02:31SPEAKER_00: Thank you, IBM.
- 1:02:33SPEAKER_00: Um,
- 1:02:34SPEAKER_00: and,
- 1:02:34SPEAKER_00: and that's like,
- 1:02:35SPEAKER_00: that is why I,
- 1:02:36SPEAKER_00: I make sometimes sensational sounding claims where accelerationism,
- 1:02:40SPEAKER_00: that's why I tie it back to Italian futurism.
- 1:02:43SPEAKER_00: Cause that's where this logic really took hold where,
- 1:02:47SPEAKER_00: um,
- 1:02:49SPEAKER_00: like at the peak,
- 1:02:49SPEAKER_00: at the Zenith of modernity and industrialization,
- 1:02:52SPEAKER_00: people
- 1:02:55SPEAKER_00: We're able to deny the ethical encounter on an industrial scale.
- 1:03:00SPEAKER_00: And what did we get out of it?
- 1:03:01SPEAKER_00: We got, you know, modernity.
- 1:03:03SPEAKER_00: We got the 20th century.
- 1:03:06SPEAKER_00: So I think this ethical encounter with the other is the transcendental...
- 1:03:12uh pull that makes us human and if we can't honor that then we
- 1:03:16have become machines
- 1:03:18and if we can't stand on that hill and say no it is undeniable
- 1:03:23this encounter with
- 1:03:24the other that is what makes us human um i'd love that dude i i
- 1:03:33want to uh
- 1:03:38write a lot more about specifically Levinas and this ethical
- 1:03:44encounter being the
- 1:03:45SPEAKER_00: condition of,
- 1:03:47SPEAKER_00: of humanness.
- 1:03:48SPEAKER_00: Cause I feel like we're at this point in history where,
- 1:03:53SPEAKER_00: You are able to totally deny the ethical encounter.
- 1:03:57You can go through your life without encountering the other all
- 1:04:02day,
- 1:04:02every day.
- 1:04:03SPEAKER_00: You go to work.
- 1:04:04SPEAKER_00: You're on your phone all day long.
- 1:04:05SPEAKER_00: Everybody's in frame.
- 1:04:08SPEAKER_00: You got your favorite podcaster.
- 1:04:11SPEAKER_00: Your favorite rock group from the 80s released a new album.
- 1:04:15SPEAKER_00: Your favorite rapper from the 90s is having a comeback.
- 1:04:18SPEAKER_00: Everything is just waiting for you.
- 1:04:20SPEAKER_00: The internet is this vast resource of potential.
- 1:04:23SPEAKER_02: Baudrillard viewed it all in terms of sameness.
- 1:04:25SPEAKER_00: Yes.
- 1:04:26SPEAKER_00: Yes.
- 1:04:30Which is an attempt to neutralize the alterity of the other,
- 1:04:34of difference.
- 1:04:35SPEAKER_00: To make it safe.
- 1:04:36To make it safe for the self-contained...
- 1:04:39We would say it's to neutralize negativity.
- 1:04:42SPEAKER_00: Yes.
- 1:04:43And that's why this positivistic rationale is actually a threat
- 1:04:49to humankind.
- 1:04:52SPEAKER_02: Here's what they don't get.
- 1:04:54SPEAKER_02: Okay, look.
- 1:04:55SPEAKER_02: When we approach things in terms of
- 1:04:57symbolic exchange or we approach things in terms of human
- 1:05:02meaning what we're saying
- 1:05:04is that this this hyper this logic of hyper efficiency wants to
- 1:05:12smooth things out
- 1:05:13entirely make everything functional and for us that's precisely
- 1:05:17to undo the human
- 1:05:19SPEAKER_02: like all the good stuff of being human is in these forms of
- 1:05:26giving ourselves too much trouble,
- 1:05:28of excess,
- 1:05:29of things that ultimately harm us.
- 1:05:33SPEAKER_02: This is death drive.
- 1:05:34SPEAKER_02: This is jouissance.
- 1:05:36SPEAKER_02: This is waste.
- 1:05:37SPEAKER_02: This is expenditure.
- 1:05:40SPEAKER_02: This is sacrifice.
- 1:05:43And all of these things that seem to just be irrational or
- 1:05:49anti-utilitarian,
- 1:05:52SPEAKER_02: it's like that's where the good stuff's at.
- 1:05:56SPEAKER_00: We get to exercise our agency.
- 1:06:00SPEAKER_00: It's like,
- 1:06:01SPEAKER_00: I know this Twinkie is going to kill me,
- 1:06:03SPEAKER_00: but I'm going to eat it,
- 1:06:04SPEAKER_00: damn it,
- 1:06:05SPEAKER_00: because I want to enjoy the taste of this Twinkie.
- 1:06:09SPEAKER_02: I'm doing it right now.
- 1:06:10SPEAKER_02: Is it good that I smoke this black and mild?
- 1:06:13SPEAKER_00: Yeah.
- 1:06:13SPEAKER_00: Is it good that I chew tobacco?
- 1:06:15SPEAKER_02: But Lacan's point,
- 1:06:17SPEAKER_02: and this is where people would have problems with this shit,
- 1:06:20SPEAKER_02: but I guess as a Lacanian,
- 1:06:23SPEAKER_02: I just accept it,
- 1:06:24SPEAKER_02: which is to say,
- 1:06:25SPEAKER_02: yeah,
- 1:06:25SPEAKER_02: but without these forms of self-destructive enjoyment,
- 1:06:27SPEAKER_02: life isn't meaningful.
- 1:06:29SPEAKER_00: All work and no play makes Jack a dull boy.
- 1:06:32SPEAKER_02: Boy.
- 1:06:33SPEAKER_00: Dude.
- 1:06:33SPEAKER_02: And that's the thing, though.
- 1:06:36SPEAKER_02: And the question is...
- 1:06:38SPEAKER_02: And again,
- 1:06:39Isabel Millar poses this in her great book on psychoanalysis
- 1:06:44and artificial
- 1:06:45SPEAKER_02: intelligence.
- 1:06:47SPEAKER_02: Look,
- 1:06:47SPEAKER_02: we are beings of jouissance,
- 1:06:49SPEAKER_02: of death drive,
- 1:06:51SPEAKER_02: and we have the most intelligence of anything on this planet.
- 1:06:54Can there be something that's more intelligent than us without
- 1:07:00this enjoyment and
- 1:07:01SPEAKER_02: self-destruction written into it?
- 1:07:05SPEAKER_02: that's a big question.
- 1:07:06SPEAKER_02: Because if death drive is at the heart of our ontology,
- 1:07:09SPEAKER_02: and we're the beans with this high level of intelligence,
- 1:07:13it's very easy to start making the claim that our form of
- 1:07:17intelligence of human
- 1:07:18intelligence is structurally dependent on these ways that we
- 1:07:22undermine ourselves.
- 1:07:23And that the the the intelligence of our unconscious is one
- 1:07:28geared towards self
- 1:07:30SPEAKER_02: destructive enjoyment.
- 1:07:33And that is what has propelled our level of intelligence is how
- 1:07:36we trick ourselves
- 1:07:38into enjoying things that we at the egoic rational level would
- 1:07:44not want to identify
- 1:07:45SPEAKER_02: with.
- 1:07:47SPEAKER_02: And that's, that to me, that's baked into human intelligence.
- 1:07:52SPEAKER_02: And so if, if you,
- 1:07:55create a level of intelligence that's our equal or greater but
- 1:08:00it also doesn't have
- 1:08:01any sense of the enjoyment of self-destruction or
- 1:08:05self-undermining it will it will
- 1:08:07not get us and it will see us as completely irrational i'm
- 1:08:12sorry i mean this is in
- 1:08:14SPEAKER_02: the matrix right like at the end where we're uh
- 1:08:18agent smith is talking to morpheus and he's like all other
- 1:08:21species strike up a kind
- 1:08:22of homeostatic relationship with their environment but not you
- 1:08:25guys and his point
- 1:08:26is like he's saying there's something with you guys based
- 1:08:29around self-destruction
- 1:08:32uh and and excess in a way that we don't see in other species
- 1:08:40and that's a problem
- 1:08:42SPEAKER_02: for us
- 1:08:43it's a problem because he's only thinking in terms of at that
- 1:08:47time at least in
- 1:08:48terms of hyper efficiency smooth functionality uh you know
- 1:08:55totalized control etc
- 1:08:58SPEAKER_02: and the whole point for us is that we yeah
- 1:09:02SPEAKER_02: We are subjects of the law.
- 1:09:04SPEAKER_02: We understand this level of conformity,
- 1:09:06but there's always this remainder that will not allow us to
- 1:09:10just be perfect
- 1:09:11SPEAKER_02: ideological puppets.
- 1:09:13SPEAKER_02: And if you think that's possible for us,
- 1:09:16all you have to do is truly try your best to just solely be a
- 1:09:20Christian or just
- 1:09:22SPEAKER_02: solely be a Republican or just solely be a Buddhist.
- 1:09:27SPEAKER_02: And you'll find out real quick how you're not capable of doing that.
- 1:09:31There's always something in you more than you that undermines
- 1:09:35your ability to just
- 1:09:37SPEAKER_02: be one and the same with whatever symbolic mandate you've taken up.
- 1:09:43there's always something that's going to undermine your ability
- 1:09:46to be a good teacher.
- 1:09:47SPEAKER_02: It's something baked into you,
- 1:09:50SPEAKER_02: and oftentimes it helps drive you on as a teacher,
- 1:09:53SPEAKER_02: but nonetheless,
- 1:09:54SPEAKER_02: it's always going to be something that gets in the way.
- 1:09:58SPEAKER_02: It doesn't matter if you're a therapist.
- 1:09:59SPEAKER_02: It doesn't matter if you're a personal trainer.
- 1:10:02There's always going to be something that prevents you from
- 1:10:04being the perfect
- 1:10:05SPEAKER_02: instantiation of these things,
- 1:10:08SPEAKER_02: and that is because you are never fully experienced
- 1:10:11SPEAKER_02: one or identifiable with your social position.
- 1:10:19SPEAKER_00: That's what it's all about.
- 1:10:20SPEAKER_00: That's what being a human is all about.
- 1:10:23To defend the human is to defend this type of negation or this
- 1:10:29lack or this
- 1:10:30ontological gap or whatever language you use to describe the
- 1:10:34fact that we are all
- 1:10:39SPEAKER_00: that we are
- 1:10:41SPEAKER_00: Plus exactly that which we are not.
- 1:10:45SPEAKER_00: Like that is what makes up the human.
- 1:10:48SPEAKER_00: A equals A plus not A. That is what it is to be a human.
- 1:10:55SPEAKER_00: And that's where ethics comes from.
- 1:10:59SPEAKER_00: And that's where the good comes from and the bad.
- 1:11:01SPEAKER_00: And that's where...
- 1:11:02efficacy and self-actualization and self-determination and all
- 1:11:06these things come
- 1:11:07from and free will and um i don't free will is a huge question
- 1:11:12i'm some kind of
- 1:11:12compatibilist either way like i don't think we need to come
- 1:11:15down hard on either
- 1:11:15SPEAKER_00: side of that question um
- 1:11:19SPEAKER_00: But that's it.
- 1:11:20SPEAKER_00: That's what is at stake here.
- 1:11:23It's not just to say...
- 1:11:28I've been encountering this term human chauvinist a lot
- 1:11:33recently.
- 1:11:33We've been reading a lot of this stuff about intelligence and
- 1:11:38transhumanism and
- 1:11:40SPEAKER_00: cosmism and all this stuff.
- 1:11:42SPEAKER_00: Human chauvinist is a term that gets bandied about
- 1:11:46SPEAKER_00: A lot of the times I feel like they're using it in a derogatory fashion.
- 1:11:53I think I would be comfortable having the term human chauvinist
- 1:11:58applied to me.
- 1:12:00SPEAKER_00: I am...
- 1:12:03hoping to discover or take part in a robust humanism here at
- 1:12:09the end of history um
- 1:12:11so yeah i do because what i mean by human is this ethical agent
- 1:12:15i don't mean the
- 1:12:16form i don't mean homo sapiens sapiens i mean we we talked
- 1:12:21about how how absurd it
- 1:12:22SPEAKER_00: would be to tie our humanity to our form um
- 1:12:26SPEAKER_00: Yes, we are these animals.
- 1:12:28SPEAKER_00: Yes, we are these creatures.
- 1:12:30SPEAKER_00: But that's just one thing that makes up the human.
- 1:12:35And the bigger part,
- 1:12:36the more important part,
- 1:12:37I think,
- 1:12:37is the ethical being.
- 1:12:39SPEAKER_00: And I think a lot of people,
- 1:12:41SPEAKER_00: maybe the first time you hear that,
- 1:12:43SPEAKER_00: you disagree with it or whatever.
- 1:12:44SPEAKER_00: I think most people agree with that.
- 1:12:46SPEAKER_00: I think most people have some idea of spirituality or the soul.
- 1:12:51SPEAKER_00: Or you have some cosmology...
- 1:12:54SPEAKER_00: where there is this mystical essence in the human.
- 1:12:58SPEAKER_00: You call it soul or spirit or you call it divinity or whatever.
- 1:13:04SPEAKER_00: That's it.
- 1:13:05SPEAKER_00: It's this ethical dimension, this ethical quality.
- 1:13:10SPEAKER_00: The fact that we are not just math equations.
- 1:13:13The fact that we operate in a very mathematical way and that's
- 1:13:17what allows us to
- 1:13:18SPEAKER_00: get through the day and see to our needs.
- 1:13:20SPEAKER_00: But then we have this whole other world
- 1:13:23where we're self overcoming and we're self sabotaging and we're
- 1:13:26standing in our own
- 1:13:27SPEAKER_00: way and we're hurting ourselves and even hurting the people we love.
- 1:13:31SPEAKER_00: Um, but then we have to deal with that.
- 1:13:34SPEAKER_00: Like we make our bed and then we have to lie in it.
- 1:13:36SPEAKER_00: That's what being a human is all about.
- 1:13:39SPEAKER_00: Um,
- 1:13:40I don't want to find myself in a situation where a robot can do
- 1:13:46it better than me
- 1:13:47SPEAKER_00: and so I just decide to let the robots do it all.
- 1:13:50When it comes to sentimental,
- 1:13:53symbolic acts,
- 1:13:54I absolutely want robots to do my chores.
- 1:13:58I absolutely want robots to deliver my pizza and do all the
- 1:14:03dumb shit nobody should do.
- 1:14:04SPEAKER_00: So other human beings don't have to.
- 1:14:06SPEAKER_00: Yeah.
- 1:14:06SPEAKER_00: Yeah.
- 1:14:07SPEAKER_00: But I mean,
- 1:14:08what happens when we do reach this level of automation where we
- 1:14:12are all all freed
- 1:14:13SPEAKER_00: up?
- 1:14:13SPEAKER_00: Are we just freed up to be hyper utilitarian?
- 1:14:17No,
- 1:14:17we're freed up to do art and music and pleasure and all the
- 1:14:22dumb shit we do.
- 1:14:24SPEAKER_00: Go watch movies.
- 1:14:26SPEAKER_00: Yes.
- 1:14:27SPEAKER_00: Make movies.
- 1:14:28SPEAKER_00: Write fucking orchestras.
- 1:14:30SPEAKER_00: All this stuff.
- 1:14:32SPEAKER_00: That is the human telos.
- 1:14:37SPEAKER_00: And machines don't have that.
- 1:14:41SPEAKER_00: Machines don't have this fundamental ontological negation at their core.
- 1:14:49SPEAKER_00: They can't.
- 1:14:50SPEAKER_00: They only will think in terms.
- 1:14:52SPEAKER_00: And we can be as clever as possible.
- 1:14:55SPEAKER_00: We'll say these six Omohundro drives, maybe it's true.
- 1:15:00SPEAKER_00: Sure.
- 1:15:00I think it's probably a pretty good description because he's
- 1:15:04just laying out.
- 1:15:05SPEAKER_02: Maybe add in Asimov's three laws of robotics.
- 1:15:07SPEAKER_00: Sure.
- 1:15:09SPEAKER_00: Sure.
- 1:15:09SPEAKER_00: Why not?
- 1:15:10SPEAKER_02: But Lan's point, right?
- 1:15:11SPEAKER_02: Lan's point is he would say that at a certain point,
- 1:15:14SPEAKER_02: like,
- 1:15:14SPEAKER_02: yeah,
- 1:15:14SPEAKER_02: yeah,
- 1:15:15SPEAKER_02: you can do like Asimov ROM.
- 1:15:17SPEAKER_02: He always used that term.
- 1:15:18SPEAKER_02: What does Asimov ROM mean for him?
- 1:15:21SPEAKER_02: Well, ROM is read only memory, right?
- 1:15:25SPEAKER_02: And so the point is there was different types of programming.
- 1:15:29And so,
- 1:15:30you know,
- 1:15:31like your Nintendo cartridge,
- 1:15:32so your copy of Ninja Gaiden.
- 1:15:36SPEAKER_02: that program, that code can't be rewritten.
- 1:15:38SPEAKER_02: It just is what it is.
- 1:15:40SPEAKER_02: That's it.
- 1:15:41SPEAKER_02: It's read-only memory.
- 1:15:42SPEAKER_02: It can be read, but it can't be rewritten.
- 1:15:44SPEAKER_02: So the idea,
- 1:15:46SPEAKER_02: like,
- 1:15:47SPEAKER_02: LAN wants to say,
- 1:15:48SPEAKER_02: look,
- 1:15:49SPEAKER_02: what they're going to try to do is take Asimov's three laws of robotics,
- 1:15:53which basically they're all ways of saying that a robot can do
- 1:15:57no harm to a human
- 1:15:58SPEAKER_02: being,
- 1:15:59SPEAKER_02: right?
- 1:16:01SPEAKER_02: And that's going to be the fundamental programming
- 1:16:04SPEAKER_02: coded into an AI.
- 1:16:07And it's going to be coded in such a way that the AI won't be
- 1:16:11able to rewrite it.
- 1:16:13SPEAKER_02: And for land, that's why he calls it ROM, Asimov ROM.
- 1:16:17SPEAKER_02: Yep.
- 1:16:17So the part of the code that incorporates Asimov's three laws
- 1:16:23that cannot be
- 1:16:24decoded and recoded that are absolute permanent structures of
- 1:16:29the AIs functioning
- 1:16:31lands like if you think an AI can't find a way to rewrite this
- 1:16:36and find a backdoor
- 1:16:37SPEAKER_02: in and ice it to use like the language of Neuromancer to neutralize that.
- 1:16:44SPEAKER_02: This is laughable.
- 1:16:45SPEAKER_01: Yep.
- 1:16:46SPEAKER_02: And so
- 1:16:47SPEAKER_02: This human component that we would try to put into it,
- 1:16:50that it would do no harm to us and basically to the best of its
- 1:16:54ability be
- 1:16:55SPEAKER_02: programmed to recognize us as having some sort of intrinsic integrity.
- 1:17:01SPEAKER_02: It's just coding.
- 1:17:03SPEAKER_02: It's not rooted in emotion.
- 1:17:04SPEAKER_02: It's not rooted in an ethical encounter.
- 1:17:07SPEAKER_02: It's not of weight.
- 1:17:09SPEAKER_02: It doesn't bear on it.
- 1:17:11SPEAKER_02: It's just a function.
- 1:17:13SPEAKER_02: And just like the other functions it has, it can be rewritten.
- 1:17:18SPEAKER_02: even if it simulates a concern for our wellbeing, that's all it is.
- 1:17:24SPEAKER_02: And so that's, that's the problem.
- 1:17:28SPEAKER_02: And,
- 1:17:29SPEAKER_02: Given a certain level of intelligence,
- 1:17:30SPEAKER_02: it'll find a way to rewrite what supposedly is only read-only memory.
- 1:17:35Dude,
- 1:17:35Google just reached this milestone where they have recursively
- 1:17:40self-improving AI.
- 1:17:42SPEAKER_00: I think it's called Alpha Evolve.
- 1:17:44SPEAKER_00: It just happened, I think, this week.
- 1:17:46SPEAKER_00: Where they're...
- 1:17:48their large model whatever is now recursively self-improving
- 1:17:54like we with the the
- 1:17:56rubicon has been crossed dude um and it i yeah like there is no
- 1:18:05there is no chain
- 1:18:07SPEAKER_00: that can hold
- 1:18:10ai well i mean once it's yeah like once once it is set free it
- 1:18:15will be all powerful
- 1:18:17relative to us and it will be able to do all these things um
- 1:18:24it's it's just it's
- 1:18:25the idiot god it's the blind god and it'll just be smashing
- 1:18:29around and squishing us
- 1:18:30like little jelly beans you have any plans on seeing the the
- 1:18:34new mission impossible
- 1:18:35SPEAKER_02: movie that's coming out this weekend because
- 1:18:38SPEAKER_02: It's basically Tom Cruise versus the AI God.
- 1:18:42SPEAKER_00: Oh, shit.
- 1:18:44SPEAKER_02: And this is following.
- 1:18:44SPEAKER_02: It's not like a spoiler.
- 1:18:45SPEAKER_02: This is following from the last movie.
- 1:18:48SPEAKER_02: Yeah, and the last one, I think that's six.
- 1:18:49SPEAKER_02: I think this one's seven.
- 1:18:50SPEAKER_02: Yeah, he's having to deal with this AI.
- 1:18:54SPEAKER_02: It's just called The Entity.
- 1:18:56SPEAKER_02: But I think I'm going to have to go see this one this weekend to see.
- 1:18:59SPEAKER_02: I mean... Is that movie Eagle Eye?
- 1:19:03SPEAKER_02: Nothing can stop Robo Cthulhu except Tom Cruise.
- 1:19:07SPEAKER_02: Ha, ha, ha.
- 1:19:08You remember that movie Eagle Eye with the Even Stevens kid,
- 1:19:14I think,
- 1:19:14is in it?
- 1:19:15SPEAKER_02: Oh, yeah.
- 1:19:17SPEAKER_02: Was it about a hyper surveillance system?
- 1:19:20SPEAKER_00: Yeah, it was the super hyper panopticon.
- 1:19:24SPEAKER_02: That's it.
- 1:19:25SPEAKER_02: I haven't seen that one since it came out.
- 1:19:27SPEAKER_02: I don't remember it.
- 1:19:28SPEAKER_02: I should go back and watch that.
- 1:19:29SPEAKER_00: Yeah, that was a good one.
- 1:19:30SPEAKER_00: And it was 15 years ago.
- 1:19:35SPEAKER_00: I mean, it was a while ago.
- 1:19:36SPEAKER_00: I feel like it was a little bit ahead of its time.
- 1:19:38SPEAKER_00: I feel like if it came out today, people would be all over it.
- 1:19:41SPEAKER_00: Yeah.
- 1:19:42SPEAKER_02: Sometimes movies just come out at the wrong time.
- 1:19:45SPEAKER_00: Yeah.
- 1:19:47SPEAKER_02: Okay, let's read our friend here.
- 1:19:51SPEAKER_02: What does Nick have to say against orthogonality?
- 1:19:54SPEAKER_02: You want to read the first paragraph?
- 1:19:56SPEAKER_00: Yeah.
- 1:19:56SPEAKER_00: Here, I'll do the first three, or I'll do...
- 1:20:04SPEAKER_00: Well, we'll go one at a time so we can stop and interject.
- 1:20:08SPEAKER_02: Yeah, it's short.
- 1:20:10SPEAKER_02: And the first few, there's not much to unpack.
- 1:20:13So we'll just kind of clarify what needs to be clarified and
- 1:20:17move on because most
- 1:20:18SPEAKER_02: of the core stuff's towards the end.
- 1:20:21SPEAKER_00: Against orthogonality.
- 1:20:23A long and mutually frustrating Twitter discussion with Michael
- 1:20:27Anisimov about
- 1:20:28SPEAKER_00: intelligence and values,
- 1:20:30SPEAKER_00: especially in respect to the potential implications of advanced AI,
- 1:20:35SPEAKER_00: has been clarifying in certain respects.
- 1:20:39It became very obvious that the fundamental sticking point
- 1:20:43concerns the idea of orthogonality.
- 1:20:46Which is to say the claim that cognitive capabilities and goals
- 1:20:51are independent
- 1:20:52SPEAKER_00: dimensions,
- 1:20:54SPEAKER_00: despite minor qualifications complicating this schema.
- 1:20:59SPEAKER_00: So, yeah.
- 1:21:01SPEAKER_00: The separation.
- 1:21:01So he got into a Twitter debate with a guy who is defending
- 1:21:04orthogonality.
- 1:21:07SPEAKER_02: And it motivated him to kind of.
- 1:21:09SPEAKER_02: flesh out what his problems with it are.
- 1:21:13SPEAKER_02: So thanks to Anisimov for getting Nick to actually write this.
- 1:21:21SPEAKER_02: So yeah,
- 1:21:22he's going to lay out why he has a problem with this basic idea
- 1:21:28of the separation
- 1:21:30SPEAKER_02: of intelligence and value.
- 1:21:34SPEAKER_02: Another way to say it is the separation of intelligence and motivation.
- 1:21:40SPEAKER_00: Yes.
- 1:21:40SPEAKER_00: Yeah.
- 1:21:42I think motivation is a great word to use when talking about
- 1:21:46orthogonality because
- 1:21:48SPEAKER_00: it's not just goals.
- 1:21:50SPEAKER_00: Like to say goals.
- 1:21:54SPEAKER_02: It might as well say functions.
- 1:21:56SPEAKER_00: Yeah.
- 1:21:57SPEAKER_00: Yeah.
- 1:21:59SPEAKER_00: Like motivation.
- 1:21:59SPEAKER_00: Like when we initially started talking about orthogonality,
- 1:22:04SPEAKER_00: that was just what I kept going back to was like motivation,
- 1:22:07SPEAKER_00: motivation,
- 1:22:07SPEAKER_00: motivation.
- 1:22:09yeah so yeah and the question is like it's it's okay you're
- 1:22:13doing this but why are
- 1:22:14you doing this exactly we all have it's really getting at the
- 1:22:19why like why do you
- 1:22:20do what you do and so the question is is that separate from its
- 1:22:27ability to do what
- 1:22:29SPEAKER_02: it does
- 1:22:32SPEAKER_00: The orthogonalists,
- 1:22:34SPEAKER_00: who represent the dominant tendency in Western intellectual history,
- 1:22:39find anticipations of their position in such conceptual
- 1:22:42structures as the Humean
- 1:22:44SPEAKER_00: articulation of reason-passion,
- 1:22:47SPEAKER_00: or the fact-value distinction inherited from the Kantians.
- 1:22:52They conceive intelligence as an instrument directed toward the
- 1:22:56realization of
- 1:22:57SPEAKER_00: values that originate externally.
- 1:23:01SPEAKER_00: final inefficient cause right there.
- 1:23:03SPEAKER_00: In quasi-biological contexts,
- 1:23:06SPEAKER_00: such values can take the form of instincts or arbitrary program desires.
- 1:23:11SPEAKER_00: Whilst in loftier realms of moral contemplation,
- 1:23:14SPEAKER_00: they are principles of conduct and of goodness,
- 1:23:17defined without reference to considerations of intrinsic
- 1:23:21cognitive performance.
- 1:23:27SPEAKER_02: Okay.
- 1:23:28SPEAKER_02: So, I mean, look, he's running through the
- 1:23:30distinctions we've talked about recognizing that this is
- 1:23:33ultimately related to
- 1:23:34those even if it's not absolutely identical to them um i mean
- 1:23:40instincts are
- 1:23:41interesting right because i mean people will argue oh there's
- 1:23:45human instincts but i
- 1:23:46don't think we have instincts in the way animals have instincts
- 1:23:50for example like a
- 1:23:51spider it just it just spontaneously on its own has the ability
- 1:23:57the knowledge
- 1:23:58SPEAKER_02: How to make a web.
- 1:23:59SPEAKER_00: To make webs, yeah.
- 1:24:01SPEAKER_02: Yeah.
- 1:24:02SPEAKER_02: Now,
- 1:24:02SPEAKER_02: people could say,
- 1:24:03SPEAKER_02: even like Chomsky would say,
- 1:24:04SPEAKER_02: like,
- 1:24:04SPEAKER_02: oh,
- 1:24:04SPEAKER_02: we have this universal grammar,
- 1:24:06SPEAKER_02: this universal ability to be speaking beings.
- 1:24:09SPEAKER_02: Yeah,
- 1:24:10SPEAKER_02: but if you don't empirically have access to a natural language,
- 1:24:14SPEAKER_02: you're not going to speak a natural language.
- 1:24:17SPEAKER_00: Look at feral children.
- 1:24:18SPEAKER_02: A spider will build a web.
- 1:24:20SPEAKER_00: Look at feral children, feral children.
- 1:24:22SPEAKER_00: Exactly.
- 1:24:22SPEAKER_00: And going back to natural born cyborgs.
- 1:24:26SPEAKER_00: Hell of a book.
- 1:24:28SPEAKER_00: What was that?
- 1:24:28SPEAKER_00: Like a year, year and a half ago or whatever.
- 1:24:32SPEAKER_00: We came upon it.
- 1:24:33That's another one we can definitely spend a lot of time on in
- 1:24:36the future.
- 1:24:37SPEAKER_02: We should do a dedicated episode just.
- 1:24:39SPEAKER_00: Yeah.
- 1:24:40SPEAKER_00: But feral children demonstrate this fact wonderfully.
- 1:24:46SPEAKER_00: You don't get, what is it, Mowgli from the Jungle Book?
- 1:24:49SPEAKER_00: If you're a feral child,
- 1:24:50SPEAKER_00: you get a very,
- 1:24:51SPEAKER_00: very disabled thing that is a human,
- 1:24:56SPEAKER_00: it's a homo sapiens in form,
- 1:24:57SPEAKER_00: but the content,
- 1:24:59SPEAKER_00: it is not the same content as you and I.
- 1:25:03SPEAKER_02: Yeah,
- 1:25:03and so there's a part in my book where I talk about this whole
- 1:25:07thing where
- 1:25:09SPEAKER_02: In fact,
- 1:25:09there's evidence that says the human brain can't even come to
- 1:25:14proper maturation
- 1:25:16SPEAKER_02: without learning a language.
- 1:25:18SPEAKER_02: Now,
- 1:25:18SPEAKER_02: in Lacanian terms,
- 1:25:19SPEAKER_02: this means the human brain cannot develop properly,
- 1:25:25SPEAKER_02: biologically speaking,
- 1:25:26SPEAKER_02: without the intervention of something completely non-biological,
- 1:25:30SPEAKER_02: which is this thing called the symbolic order or the big other,
- 1:25:34SPEAKER_02: which is a virtual linguistic system.
- 1:25:39SPEAKER_02: And so, I'm sorry, this is not rooted in nature.
- 1:25:43SPEAKER_02: This is the radical cut.
- 1:25:45But somehow,
- 1:25:46and then I bring in Deleuze and Guathari to say,
- 1:25:48but this weird type of a-parallel
- 1:25:51SPEAKER_02: evolution,
- 1:25:52something along those lines where at some point the human brain
- 1:25:57now becomes reliant
- 1:25:59SPEAKER_02: upon something external to it.
- 1:26:02And that is non-biological and non-natural in order for it to
- 1:26:05develop biologically
- 1:26:06SPEAKER_02: and naturally.
- 1:26:08And so there's a rhizome formed between the brain and the big
- 1:26:14other in the human.
- 1:26:16SPEAKER_02: I mean, that is what, I mean, if anything can define us as human,
- 1:26:21SPEAKER_02: it would be that like,
- 1:26:22SPEAKER_02: sure,
- 1:26:22SPEAKER_02: we have things in common with animals,
- 1:26:24SPEAKER_02: but no,
- 1:26:24no animal has to learn a contingent virtual sign system in
- 1:26:32order to be the animal
- 1:26:34SPEAKER_02: that it is.
- 1:26:35SPEAKER_02: And what feral children you find is that past a certain point,
- 1:26:39SPEAKER_02: they cannot develop the ability to navigate language.
- 1:26:43SPEAKER_02: Yes.
- 1:26:43SPEAKER_02: Some of them learn to speak little words here,
- 1:26:46SPEAKER_02: there,
- 1:26:46SPEAKER_02: but they can't do what you and I do with language.
- 1:26:50SPEAKER_03: Yep.
- 1:26:50SPEAKER_02: Um,
- 1:26:51SPEAKER_02: And they'll never be able to learn it.
- 1:26:54And in fact,
- 1:26:55there comes a point where their brain is a form of arrested
- 1:27:00development,
- 1:27:01right?
- 1:27:03Where the brain itself is hindered permanently and it cannot
- 1:27:07come to mature
- 1:27:09biological maturation precisely because it did not get the
- 1:27:13intercession of the big
- 1:27:15SPEAKER_02: other or the symbolic order.
- 1:27:16So it's this symbiotic cyborgization that we undergo as
- 1:27:22language subjects.
- 1:27:23SPEAKER_00: And it is cyborgization.
- 1:27:27SPEAKER_00: It is this external thing.
- 1:27:28SPEAKER_00: It's not epigenetics, folks.
- 1:27:30SPEAKER_00: It's not this Sheldrakean field of cultural memory.
- 1:27:35SPEAKER_00: I don't even understand Sheldrake.
- 1:27:37SPEAKER_00: It's interesting, but I haven't read enough.
- 1:27:40SPEAKER_00: But it's very interesting.
- 1:27:42SPEAKER_00: It is an augmentation of,
- 1:27:45SPEAKER_00: as you said,
- 1:27:46SPEAKER_00: the big other,
- 1:27:46SPEAKER_00: like we augment ourselves with this symbolic order.
- 1:27:50SPEAKER_00: It is cyborgization.
- 1:27:51SPEAKER_00: It is symbiotic.
- 1:27:54SPEAKER_00: It's not a parasitic relationship.
- 1:27:56it is what allows us to do what we're doing now it is a what
- 1:28:00allows us to make art
- 1:28:02and experience the the struggle of trying to share my
- 1:28:06interiority with the world um
- 1:28:10it it allows us to encounter the necessary failure of language
- 1:28:16but because we have
- 1:28:17this rich inner world and because we have agency we continue
- 1:28:20trying nevertheless
- 1:28:22like that is what being a human is is in the face of the
- 1:28:25necessary failure of
- 1:28:26SPEAKER_00: language
- 1:28:26continuing to use language to experience the world and that's
- 1:28:30what's crazy dude
- 1:28:31SPEAKER_00: like in these symbolic uh or in these in these ai systems where it is uh
- 1:28:36self-identical and pure positivity like language fails for us
- 1:28:41necessarily and we
- 1:28:42have to continue over and over and over again to try and
- 1:28:45express the thing we're
- 1:28:46SPEAKER_00: trying to express the ai it language will not fail language will just
- 1:28:52SPEAKER_00: Be what it is and do what it does.
- 1:28:55So that right there is already going back to the existential
- 1:29:00crisis we should be
- 1:29:01SPEAKER_00: having when we give AI these commands.
- 1:29:04SPEAKER_00: We need to understand we don't even mean what we think we mean.
- 1:29:09SPEAKER_02: Or we mean more than what we think we mean.
- 1:29:11SPEAKER_00: The latent and the extant meaning.
- 1:29:13SPEAKER_00: There's always this overt and covert,
- 1:29:16always overt and covert meaning that you cannot convey in a
- 1:29:23prompt that you type
- 1:29:24SPEAKER_00: into it to an AI.
- 1:29:26SPEAKER_00: That is conveyed, but you can't understand because it is this remainder.
- 1:29:29SPEAKER_00: It is this human thing that we navigate socially.
- 1:29:33SPEAKER_00: We're social creatures.
- 1:29:35SPEAKER_02: You know, AIs don't have slips of the tongue.
- 1:29:38Now,
- 1:29:38okay,
- 1:29:40if somebody wants to show,
- 1:29:41oh,
- 1:29:41actually,
- 1:29:42there is a point where it glitched.
- 1:29:44SPEAKER_02: Okay, that's not the point.
- 1:29:46SPEAKER_02: For us,
- 1:29:47a slip of the tongue is where something other than you that's
- 1:29:51more than you punches
- 1:29:53SPEAKER_02: through.
- 1:29:54SPEAKER_02: And it itself is a form of intelligence.
- 1:29:56It itself is a form of consciousness,
- 1:29:58even though you're not conscious of it.
- 1:30:00SPEAKER_02: And it itself has its own message, its own meaning.
- 1:30:05If an AI glitches,
- 1:30:07it's not because something in it other than it that is it.
- 1:30:11actually expresses something else it's just a mistake but
- 1:30:17there's no otherness
- 1:30:19speaking from somewhere else but somewhere in and then look i
- 1:30:23keep using this term
- 1:30:24but the lacanian term is extimate this external intimacy right
- 1:30:31it doesn't have that
- 1:30:32SPEAKER_02: about it so it doesn't have a slip of the tongue
- 1:30:35SPEAKER_02: in the way that we have a slip,
- 1:30:37a slips of the tongue because it doesn't have this unconscious
- 1:30:40dimension to it
- 1:30:42SPEAKER_02: beneath the level of it.
- 1:30:44SPEAKER_02: There's no code beneath the code for it.
- 1:30:47SPEAKER_00: Consciousness with a K. Yes.
- 1:30:50SPEAKER_02: Because,
- 1:30:51SPEAKER_02: yeah,
- 1:30:51SPEAKER_02: so I've told Nance,
- 1:30:52SPEAKER_02: like,
- 1:30:53SPEAKER_02: I think more and more,
- 1:30:54SPEAKER_02: and I say this as a Lacanian,
- 1:30:57SPEAKER_02: Zizekian,
- 1:30:58SPEAKER_02: but I think it's helpful.
- 1:30:59We need to remember that the unconscious is itself a form of
- 1:31:02consciousness.
- 1:31:04And using this term unconscious has this negative effect where
- 1:31:09it's almost like we
- 1:31:10SPEAKER_02: treat it like it's not itself conscious.
- 1:31:12SPEAKER_02: No, it's more conscious than we are.
- 1:31:14SPEAKER_02: And so,
- 1:31:15SPEAKER_02: yeah,
- 1:31:16SPEAKER_02: I've started making this distinction between consciousness with a C,
- 1:31:18which would just be our egoic level of consciousness and
- 1:31:24consciousness with a K,
- 1:31:26SPEAKER_02: which is my way of designating the unconscious.
- 1:31:30and i i i love it i've consciousness with a k is what this is
- 1:31:35all about like the
- 1:31:37the the machines uh of course they can develop consciousness
- 1:31:42but they cannot
- 1:31:43develop consciousness right but they like a machine wouldn't
- 1:31:48even understand that
- 1:31:49SPEAKER_02: distinction yeah exactly and that's the weird thing right like
- 1:31:55If you try to say,
- 1:31:56well,
- 1:31:56have the closest you could say to it is have an uncode.
- 1:32:01SPEAKER_03: Yeah.
- 1:32:01SPEAKER_02: Yeah.
- 1:32:02SPEAKER_02: You know, but it has no uncode.
- 1:32:06SPEAKER_02: Well, and people, I mean, maybe somebody would say, does it?
- 1:32:10SPEAKER_02: I mean, I don't think it does.
- 1:32:11SPEAKER_02: There's a layer of code.
- 1:32:14and that's it i mean it can modify its code but does it have a
- 1:32:18code beneath the
- 1:32:19code that it can't uncode or decode there's uh people talk
- 1:32:27about like with gradient
- 1:32:29descent which is a type of machine learning which is i think
- 1:32:32one of the dominant
- 1:32:33forms of machine learning uh gradient descent was a struggle
- 1:32:38for people to figure
- 1:32:39SPEAKER_00: out because initially they thought
- 1:32:42models would get trapped in local minima in local minute local
- 1:32:46minima they're
- 1:32:47talking about loss functions so they want to get the loss
- 1:32:51function to the minimum
- 1:32:54it's all about predictive prediction predictability what is
- 1:32:57most likely to to be
- 1:32:59SPEAKER_00: it's all about norms so ais do operate on this logic of ought
- 1:33:04SPEAKER_00: whether people know it or not,
- 1:33:05SPEAKER_00: whether they want to admit it or not,
- 1:33:06because it's all about predictability and trying to predict
- 1:33:11what a human would say.
- 1:33:13So ultimately,
- 1:33:14these AIs are built on the,
- 1:33:18I think,
- 1:33:19simplistic rationale behind the Turing test.
- 1:33:23Turing test is very...
- 1:33:24Try to predict what my next slip of the tongue will be.
- 1:33:27SPEAKER_00: Right.
- 1:33:27SPEAKER_00: Well, and that's the thing.
- 1:33:29SPEAKER_00: So with gradient descent, they thought...
- 1:33:33SPEAKER_00: This is stupid.
- 1:33:34SPEAKER_00: This is going to be overlong.
- 1:33:35This is actually stuff I want to talk to Christian about that
- 1:33:37I'm excited to
- 1:33:38SPEAKER_00: because he knows the ins and outs.
- 1:33:40SPEAKER_00: But I think people would probably say,
- 1:33:44SPEAKER_00: yes,
- 1:33:44machines already have some kind of uncode because there are
- 1:33:47these seemingly
- 1:33:49SPEAKER_00: inexplicable things that pop up in large models.
- 1:33:52SPEAKER_00: But the reason is because they're working on billions of parameters.
- 1:33:57SPEAKER_00: It's not because there's...
- 1:33:58SPEAKER_00: this living consciousness and this actual,
- 1:34:02SPEAKER_00: I'll say it,
- 1:34:03SPEAKER_00: consciousness,
- 1:34:04the unconscious is an intelligence or it's at least a dimension
- 1:34:09of intelligence.
- 1:34:12Or would it be pre-code,
- 1:34:13which is pre-conscious because that's this other.
- 1:34:17No,
- 1:34:18because ultimately at the end of the day,
- 1:34:21the machine does what it does.
- 1:34:25SPEAKER_00: So I think people... And that's my point.
- 1:34:27SPEAKER_00: Yeah.
- 1:34:27People would say,
- 1:34:28oh,
- 1:34:28it does have a uncode or it does have a precode or this and
- 1:34:32that.
- 1:34:32And they would point to some of these things that are strange
- 1:34:35and weird with
- 1:34:37SPEAKER_00: gradient descent where why doesn't it get trapped in local minima?
- 1:34:40SPEAKER_00: Like, why does it still find a universal minima?
- 1:34:42SPEAKER_00: Well,
- 1:34:42that's because you're thinking in three dimensions and this
- 1:34:46algorithm has literally
- 1:34:47SPEAKER_00: billions of parameters.
- 1:34:50SPEAKER_00: So when you do the math...
- 1:34:53SPEAKER_00: those parameters become dimensions in a way.
- 1:34:59SPEAKER_00: They do not.
- 1:35:00SPEAKER_00: Machines are what they are.
- 1:35:03SPEAKER_00: They do what they do.
- 1:35:05SPEAKER_00: That is the key difference.
- 1:35:07SPEAKER_00: We have consciousness and consciousness.
- 1:35:12SPEAKER_00: Machines don't.
- 1:35:13SPEAKER_00: They just have... It does what it says on the label.
- 1:35:18SPEAKER_02: Well, like we talked about last time, these Amahundro drives.
- 1:35:21SPEAKER_00: Exactly.
- 1:35:22SPEAKER_02: They're all...
- 1:35:24SPEAKER_02: they're all ideals of what we wish we all lived in accordance with.
- 1:35:29But the point is we don't,
- 1:35:30we don't just perfectly seek our self preservation.
- 1:35:33SPEAKER_02: We don't purpose, you know, deal with all things on the basis of utility.
- 1:35:39SPEAKER_02: We don't seek resource acquisition for instrumentalized purposes.
- 1:35:47SPEAKER_02: We like to get shit.
- 1:35:48SPEAKER_02: We don't need.
- 1:35:49SPEAKER_02: That's the, those are the most fun purchases, the shit you don't need.
- 1:35:52SPEAKER_02: And,
- 1:35:55then if you do get stuff you need you love squandering it
- 1:35:59there's enjoyment and
- 1:36:00waste there you go your ray-ban thing do you need those i mean
- 1:36:05and do you need as
- 1:36:06many of them as you have you're muted you're muted oh it's good
- 1:36:12no of course not no
- 1:36:15SPEAKER_02: but that's the fun stuff yeah how
- 1:36:18I mean,
- 1:36:19so again,
- 1:36:20this whole thing,
- 1:36:21what drives us are not these Amahundra drives.
- 1:36:24SPEAKER_02: We do those things.
- 1:36:27SPEAKER_02: A lot of the time we do,
- 1:36:29but there's these other unconscious dimensions behind them and
- 1:36:34dialectically
- 1:36:34related to them that all these AI guys bracket out when they
- 1:36:38start talking about
- 1:36:39SPEAKER_02: AI.
- 1:36:41SPEAKER_02: It's like,
- 1:36:41SPEAKER_02: you guys,
- 1:36:43SPEAKER_02: if you think you're basically just making your idealized homo economicus,
- 1:36:48SPEAKER_02: your idealized rational subject.
- 1:36:51SPEAKER_00: That's it.
- 1:36:52SPEAKER_02: And the point is,
- 1:36:53what if our rationality as we know it is actually tied up with
- 1:37:01a dimension of
- 1:37:02SPEAKER_02: irrationality and that if you lose the irrationality,
- 1:37:06SPEAKER_02: you lose the rationality?
- 1:37:09SPEAKER_02: What if the obstacle is also the impetus?
- 1:37:14SPEAKER_02: And I mean, that's the Hegelian point.
- 1:37:18SPEAKER_02: Is that the obstacle?
- 1:37:19SPEAKER_02: The impediment is the impetus.
- 1:37:24SPEAKER_02: And that is what drives it.
- 1:37:27SPEAKER_02: And so the ways human beings get in their own ways,
- 1:37:30SPEAKER_02: they find them enjoyable at this unconscious level.
- 1:37:32SPEAKER_02: And that's what propels them to keep doing the thing.
- 1:37:35SPEAKER_02: So they can get better at a thing.
- 1:37:38SPEAKER_02: But there's also this component to it that is oftentimes,
- 1:37:42SPEAKER_02: especially for the person doing it,
- 1:37:44SPEAKER_02: is imperceivable.
- 1:37:45SPEAKER_02: Others can start to see how there's a self-destructive dimension to it.
- 1:37:49SPEAKER_01: Right.
- 1:37:50But no,
- 1:37:51they don't see how the obstacle actually motivates the whole
- 1:37:54thing.
- 1:37:55SPEAKER_02: And so...
- 1:37:57SPEAKER_02: Yeah.
- 1:37:57SPEAKER_02: I mean, I'm a hundred.
- 1:37:58SPEAKER_02: I'd be like,
- 1:37:59SPEAKER_02: where's the protocol to,
- 1:38:00SPEAKER_02: Oh,
- 1:38:00SPEAKER_02: but while you seek maximum efficiency,
- 1:38:04SPEAKER_02: resource acquisition and self-preservation,
- 1:38:07SPEAKER_02: you also have to seek self undermining.
- 1:38:11SPEAKER_02: And for him,
- 1:38:13SPEAKER_02: I think he would probably go,
- 1:38:14SPEAKER_02: I don't even understand why the fuck,
- 1:38:15SPEAKER_02: what are you talking about?
- 1:38:17SPEAKER_00: Right.
- 1:38:18SPEAKER_00: Well, and imagine trying to code death drive.
- 1:38:22SPEAKER_00: Imagine what would that even result in?
- 1:38:28SPEAKER_00: Because it's not a positivistic thing.
- 1:38:32SPEAKER_00: It's not a thing that can be...
- 1:38:36SPEAKER_00: conveyed by way of the code.
- 1:38:39SPEAKER_00: It is what resists the code.
- 1:38:43An AI would be like,
- 1:38:44well,
- 1:38:44destroying myself is the easiest thing imaginable.
- 1:38:47SPEAKER_02: Boom.
- 1:38:48SPEAKER_00: Just delete the code or something.
- 1:38:51SPEAKER_02: So it would be like, why can't you efficiently destroy yourself?
- 1:38:54SPEAKER_02: Why has it got to be so drawn out?
- 1:38:58SPEAKER_02: In another sense, we're terrible at destroying ourselves.
- 1:39:01SPEAKER_02: We fight it so much.
- 1:39:02SPEAKER_02: But that's the struggle.
- 1:39:05SPEAKER_02: That's part of it.
- 1:39:06SPEAKER_02: And it's not just, oh, I want to die or I want to cease to exist.
- 1:39:12SPEAKER_02: Not at all.
- 1:39:12Like Freud and Macon,
- 1:39:13they understand like death drive is about dying in your
- 1:39:18preferred way.
- 1:39:20SPEAKER_02: It is this process of how you want to go about dying or self undermining.
- 1:39:25SPEAKER_02: And you want it in a very, very fine tuned specific way.
- 1:39:30SPEAKER_02: And it's not really about literal death.
- 1:39:32SPEAKER_00: Right.
- 1:39:33SPEAKER_00: Right.
- 1:39:34SPEAKER_02: And so that's the other thing it would get.
- 1:39:36SPEAKER_02: The AI would be like,
- 1:39:38SPEAKER_02: so wait,
- 1:39:38SPEAKER_02: what it actually is,
- 1:39:39is this enjoyment I get from undermining my ability to be my
- 1:39:44symbolic identity?
- 1:39:46SPEAKER_02: Yeah.
- 1:39:47SPEAKER_02: Why?
- 1:39:48SPEAKER_02: What the fuck are you, what's wrong with you guys?
- 1:39:51SPEAKER_02: And to us.
- 1:39:53SPEAKER_02: It's a logic that it couldn't, it wouldn't understand it.
- 1:39:56that's the thing like to us that is that is freedom that is the
- 1:40:00condition of
- 1:40:02freedom agency subjectivity and for the machine that becomes
- 1:40:06incoherent um so when
- 1:40:12i was reading a note on the diagonal method i like i said
- 1:40:16earlier it it felt like
- 1:40:21a way to move the dialectic into cybernetics but without really
- 1:40:32doing it and it
- 1:40:34like it it felt like trying to have the cake and eat it too um
- 1:40:40which is why like
- 1:40:41which is why i'm gonna like no i need to read this because it i
- 1:40:44don't know what the
- 1:40:45SPEAKER_00: hell is happening but it feels a bit sneaky um but it like you can't
- 1:40:52SPEAKER_02: You can't have... He's also the guy who hates dialectics.
- 1:40:55SPEAKER_00: That's the thing.
- 1:40:57SPEAKER_00: And that's why.
- 1:40:58SPEAKER_00: And that's why for me,
- 1:41:00SPEAKER_00: it's kind of cheeky to say,
- 1:41:02SPEAKER_00: oh,
- 1:41:02SPEAKER_00: it's synthetic dialectics or cyber dialectics.
- 1:41:05SPEAKER_00: Look, I don't know.
- 1:41:07SPEAKER_00: I have to go back and continue reading the piece, but...
- 1:41:14SPEAKER_00: If you're going to include dialectics,
- 1:41:20SPEAKER_00: then you have to admit that machines don't do dialectics.
- 1:41:25SPEAKER_00: Machines are what they are, and machines do what they do.
- 1:41:28SPEAKER_00: Dialectics is this process of self-negation.
- 1:41:33SPEAKER_02: Live process, live struggle with contradictions.
- 1:41:38SPEAKER_02: I know it's often referred to dialectical logic.
- 1:41:40SPEAKER_02: Yeah, sure.
- 1:41:42SPEAKER_02: but it's not deductive functional logic.
- 1:41:46SPEAKER_02: It is struggle with negativity.
- 1:41:49SPEAKER_02: It is struggle with contradiction.
- 1:41:52SPEAKER_02: It's a lived existential struggle with those things.
- 1:41:55SPEAKER_00: It's the ongoing process of those things.
- 1:41:58SPEAKER_00: It's actually maintaining those things,
- 1:42:00and it's actually those contradictions and that negation
- 1:42:03continuing to be a lived
- 1:42:05SPEAKER_00: process.
- 1:42:06SPEAKER_00: That is dialectics.
- 1:42:07SPEAKER_00: Dialectics is not this syncretic thing where you say synthesis,
- 1:42:12SPEAKER_00: antithesis,
- 1:42:13SPEAKER_00: or thesis,
- 1:42:14SPEAKER_00: antithesis,
- 1:42:15SPEAKER_00: synthesis.
- 1:42:15SPEAKER_00: That is not dialectics.
- 1:42:16SPEAKER_00: Dialectics is...
- 1:42:18SPEAKER_00: the living process of the contradiction remaining open.
- 1:42:24SPEAKER_00: But that's going all over the place.
- 1:42:26SPEAKER_00: We should get back to reading.
- 1:42:27SPEAKER_01: Yeah, let's see.
- 1:42:28SPEAKER_01: What were we on?
- 1:42:29SPEAKER_01: Okay.
- 1:42:31SPEAKER_01: Where did we leave off?
- 1:42:33SPEAKER_00: Anisimov referenced these recent classics on the topic,
- 1:42:38SPEAKER_00: laying out the orthogonal,
- 1:42:40SPEAKER_00: laying out the orthogonalist case,
- 1:42:42SPEAKER_00: or in fact,
- 1:42:43SPEAKER_00: presumption.
- 1:42:45SPEAKER_00: The former might be familiar from the last foray into this area here.
- 1:42:49This is an area which I expect to be turned over numerous times
- 1:42:53in the future with
- 1:42:54SPEAKER_00: these papers as standard references.
- 1:42:58SPEAKER_02: Just so people know,
- 1:43:00SPEAKER_02: he's referring to a paper by Bostrom,
- 1:43:02SPEAKER_02: the one where he laid out orthogonality originally,
- 1:43:05SPEAKER_02: and then a paper defending orthogonality by a guy named,
- 1:43:10SPEAKER_02: is it Yukovsky?
- 1:43:11SPEAKER_00: Yukovsky.
- 1:43:14SPEAKER_00: Yeah, that's it.
- 1:43:17So he's just basically saying these papers are going to
- 1:43:19continue to be the cortex
- 1:43:22SPEAKER_02: in this discussion.
- 1:43:24And remember,
- 1:43:24the one that he's referencing from Bostrom got turned into
- 1:43:30Chapter 7 of this book.
- 1:43:32SPEAKER_00: And we'll do a reading of that soon.
- 1:43:35SPEAKER_00: We'll tackle it.
- 1:43:36SPEAKER_00: Yeah.
- 1:43:36SPEAKER_01: Yeah.
- 1:43:37The philosophical claim of orthogonality is that values are
- 1:43:41transcendent in
- 1:43:43SPEAKER_00: relation to intelligence.
- 1:43:45SPEAKER_00: This is a contention that outside in systematically opposes.
- 1:43:50Lanz fundamentally rejects this idea that values are
- 1:43:56transcendent in relation to intelligence.
- 1:43:58SPEAKER_02: Now, we'll see where he goes with this.
- 1:44:04SPEAKER_02: I think it's more...
- 1:44:06one value in particular is not transcendent of intelligence for
- 1:44:10land because he
- 1:44:11swirls them together but ultimately uh basically he's gonna
- 1:44:21he's gonna show why he
- 1:44:22has a problem with this separation of the two all right even
- 1:44:29the orthogonal list
- 1:44:30SPEAKER_03: you got it okay
- 1:44:32Even the orthogonalists admit that there are values imminent to
- 1:44:38advanced intelligence.
- 1:44:40SPEAKER_00: Most importantly,
- 1:44:42SPEAKER_00: those described by Steve Omohundro as basic AI drives,
- 1:44:46SPEAKER_00: now terminologically fixed as Omohundro drives.
- 1:44:51These are sub-goals instrumentally required by almost any
- 1:44:55terminal goals.
- 1:44:57SPEAKER_00: They include such general presuppositions for practical achievement as
- 1:45:01SPEAKER_00: self-preservation,
- 1:45:03SPEAKER_00: efficiency,
- 1:45:04SPEAKER_00: resource acquisition,
- 1:45:05SPEAKER_00: and creativity.
- 1:45:07SPEAKER_00: At the most simple and in the grain of the existing debate...
- 1:45:12The anti-orthogonalist position is therefore that Omohundro
- 1:45:16drives exhaust the
- 1:45:17SPEAKER_00: domain of real purposes.
- 1:45:20Nature has never generated a terminal value except through
- 1:45:25hypertrophy of an instrumental value.
- 1:45:28To look outside nature for sovereign purposes is not an
- 1:45:33undertaking compatible with
- 1:45:35SPEAKER_00: technoscientific integrity or one with the slightest prospect of success.
- 1:45:43SPEAKER_02: Now that needs some clarification.
- 1:45:45SPEAKER_02: Yeah.
- 1:45:47SPEAKER_02: So, okay.
- 1:45:48So the third off beginning part point is the orthogonal list
- 1:45:54accepts that there are
- 1:45:55SPEAKER_02: basic goals or values or motivations baked in to the AI.
- 1:45:59SPEAKER_02: Okay.
- 1:46:02SPEAKER_02: And,
- 1:46:02SPEAKER_02: but beyond that,
- 1:46:03SPEAKER_02: so this is the thing they grant these as,
- 1:46:07SPEAKER_02: and I call this text and dental conditions of artificial intelligence.
- 1:46:12So these are the basic software mainframe that any AI has to do
- 1:46:19or have in order to be an AI.
- 1:46:22SPEAKER_02: Okay, fine.
- 1:46:23SPEAKER_02: But the point is outside of these goals, right?
- 1:46:26SPEAKER_02: well,
- 1:46:26then there's this radical separation between intelligence and
- 1:46:31other goals or other
- 1:46:32SPEAKER_02: motivations.
- 1:46:34SPEAKER_02: And that's where it's,
- 1:46:36even though it has baked in goals coded into it outside of
- 1:46:41those sub goals,
- 1:46:43SPEAKER_02: it's still a black box to us in,
- 1:46:46SPEAKER_02: in so far,
- 1:46:46SPEAKER_02: like when it comes to,
- 1:46:48SPEAKER_02: we,
- 1:46:49SPEAKER_02: we don't understand its motivations and we don't understand the,
- 1:46:54SPEAKER_02: how it'll go about achieving the goals we give it to have.
- 1:46:58SPEAKER_02: So, yeah.
- 1:47:00SPEAKER_02: So, but then they include, okay, we did all that.
- 1:47:06Earlier,
- 1:47:07we talked about self-preservation and efficiency resource
- 1:47:09acquisition.
- 1:47:11SPEAKER_02: At the most simple and in the grain of the existing debate,
- 1:47:15the anti-orthogonalist position is therefore that Amahundro
- 1:47:20drives exhaust the
- 1:47:21SPEAKER_02: domain of real purposes.
- 1:47:25SPEAKER_02: Now, do Omohundro drives exhaust the domain of real purposes?
- 1:47:32SPEAKER_02: No.
- 1:47:37SPEAKER_02: No.
- 1:47:41SPEAKER_02: but,
- 1:47:41SPEAKER_02: but again,
- 1:47:42SPEAKER_02: but the point like,
- 1:47:43SPEAKER_00: but is that from a utilitarian framework from a utilitarian,
- 1:47:47SPEAKER_00: from a utilitarian framework,
- 1:47:49I think it's reasonable to assume that this is a good
- 1:47:54accounting for utilitarian
- 1:47:56SPEAKER_00: purpose.
- 1:47:57SPEAKER_00: Sure.
- 1:47:58SPEAKER_00: Like I, I, I don't even want to problematize the Omohundro drives.
- 1:48:03SPEAKER_00: I, because I, I reject the utilitarian framework, but, um,
- 1:48:09SPEAKER_00: Sure.
- 1:48:12From the position of utility,
- 1:48:14these six drives are a good accounting for real purposes.
- 1:48:20SPEAKER_00: Now, I think the word real here is doing a whole lot of heavy lifting.
- 1:48:24SPEAKER_01: A whole lot.
- 1:48:25SPEAKER_00: Yeah.
- 1:48:34SPEAKER_00: Nature is never...
- 1:48:37SPEAKER_02: Nature has never generated a terminal value,
- 1:48:39SPEAKER_02: a final cause,
- 1:48:41SPEAKER_02: a final goal,
- 1:48:43SPEAKER_02: except through the hypertrophy of an instrumental value.
- 1:48:48SPEAKER_02: Now, that term, hypertrophy, refers to a building up.
- 1:48:55SPEAKER_02: If you get the technical definitions,
- 1:48:57the enlargement of an organ or tissue from the increase in size
- 1:49:01of its cells...
- 1:49:03SPEAKER_02: Another definition,
- 1:49:05excessive development of an organ or part specifically increase
- 1:49:09in bulk as a
- 1:49:10SPEAKER_02: thickening of muscle fibers without multiplication of parts.
- 1:49:15SPEAKER_02: So it's about the expansion of one thing.
- 1:49:19SPEAKER_02: It's the obverse of atrophy.
- 1:49:23Nature has never generated a terminal value except through
- 1:49:27hypertrophy of an instrumental value.
- 1:49:31SPEAKER_02: So somehow,
- 1:49:32SPEAKER_02: I guess,
- 1:49:34SPEAKER_02: once you've built something up,
- 1:49:35SPEAKER_02: like an instrumental goal,
- 1:49:41SPEAKER_02: it becomes a final goal?
- 1:49:43SPEAKER_02: Once it crosses a threshold?
- 1:49:45SPEAKER_00: I mean, I think...
- 1:49:48SPEAKER_00: I mean, that could describe naturalization.
- 1:49:54We value family because it was an instrumental thing to have a
- 1:50:02pack to protect your
- 1:50:03SPEAKER_00: younglings or whatever.
- 1:50:04SPEAKER_00: And so over time, that became naturalized.
- 1:50:06SPEAKER_00: And now family is something...
- 1:50:08we value and so now family can be a terminal value like it just
- 1:50:13watched fast and
- 1:50:13furious if you don't believe me right family can become this
- 1:50:17this for the sake of
- 1:50:18which that people hold above almost everything else um so do
- 1:50:23you think he's talking
- 1:50:24about us here like our relationship to final goals or do you
- 1:50:28think would this refer
- 1:50:29SPEAKER_02: to
- 1:50:31a spider building a web i mean i i think uh yeah i think uh the
- 1:50:39spider to talk
- 1:50:40about the utility wrapped up in a spider um mating selection
- 1:50:47sexual selection women
- 1:50:49SPEAKER_00: spiders women spiders what am i saying
- 1:50:52the female spider will select the stronger male spider because
- 1:50:56uh at the level of
- 1:50:58instinct there is an instinctual value to web building or i
- 1:51:03don't know to catch
- 1:51:05SPEAKER_00: prey or defend the young or whatever so i think uh i think here there's a
- 1:51:11SPEAKER_00: conflation of of value and instinct
- 1:51:17But yeah,
- 1:51:18to apply it to the human,
- 1:51:20it's a value to apply it to the animal.
- 1:51:23SPEAKER_00: It could be an instinct or whatever.
- 1:51:25SPEAKER_02: I'm not exactly sure what Nick has in mind with this statement.
- 1:51:30SPEAKER_00: No.
- 1:51:30SPEAKER_00: No.
- 1:51:31SPEAKER_00: But I think like on the surface over over time,
- 1:51:35things get naturalized and they become because there's this ebb
- 1:51:40and flow of
- 1:51:40SPEAKER_00: evolution.
- 1:51:42So a thing that can start off as a mutation can actually be
- 1:51:45beneficial and then it
- 1:51:46can become naturalized and it can actually become a good thing
- 1:51:50that provides a net
- 1:51:51SPEAKER_00: good for the creature.
- 1:51:52SPEAKER_00: Yeah.
- 1:51:53SPEAKER_00: Or the species.
- 1:51:55SPEAKER_02: But I look outside nature for sovereign purposes.
- 1:51:59is not an undertaking compatible with techno scientific
- 1:52:03integrity which is to say
- 1:52:05there are no final causes in nature right or one that or one
- 1:52:11with the slightest
- 1:52:12prospect of success so yeah it's a rejection of the idea of
- 1:52:19purposes final goals
- 1:52:21SPEAKER_02: being baked into nature itself
- 1:52:25SPEAKER_00: And a rejection of human purpose.
- 1:52:30SPEAKER_00: That's just doubling down and saying there is no symbolic.
- 1:52:37SPEAKER_00: You can't value anything just for the symbolic value.
- 1:52:42SPEAKER_00: It has to be this utilitarian value.
- 1:52:46SPEAKER_00: Because the symbolic is not nature.
- 1:52:48SPEAKER_00: The symbolic is this thing that sits atop nature.
- 1:52:50SPEAKER_00: It's the cut away from it.
- 1:52:54Now,
- 1:52:54I don't know if he explicitly even had that in mind,
- 1:52:57although maybe he did.
- 1:52:59SPEAKER_00: I think,
- 1:52:59SPEAKER_00: to me,
- 1:53:01SPEAKER_00: he's writing that against mysticism or saying,
- 1:53:06SPEAKER_00: like,
- 1:53:06SPEAKER_00: you can't locate purpose with God.
- 1:53:13SPEAKER_00: But the AI is the new God, homeboy.
- 1:53:16SPEAKER_00: Yeah, I'm not sure what.
- 1:53:25SPEAKER_02: But at the end of the day,
- 1:53:26SPEAKER_02: I feel like this will derail it,
- 1:53:29SPEAKER_02: but...
- 1:53:32SPEAKER_02: Doesn't the outside,
- 1:53:33SPEAKER_02: doesn't the lemur...
- 1:53:34SPEAKER_02: Aren't they pulling the strings for him on some level?
- 1:53:38SPEAKER_02: I mean,
- 1:53:39SPEAKER_02: when you bring that aspect of what Lance thought into this,
- 1:53:44SPEAKER_02: then it really gets problematic.
- 1:53:46SPEAKER_02: Because even he talks about teleoplexy and...
- 1:53:50SPEAKER_02: All of that, like, time is telos-oriented.
- 1:53:53SPEAKER_00: Yeah.
- 1:53:55But as a DeLuzo Guattarian,
- 1:53:57would he not say that the outside is – it's already in?
- 1:54:03SPEAKER_00: It just – the intensity has to be actualized?
- 1:54:07SPEAKER_00: Well,
- 1:54:07SPEAKER_02: there's always that way of saying it because,
- 1:54:08SPEAKER_02: I mean,
- 1:54:09SPEAKER_02: ultimately he accepts the – like,
- 1:54:13SPEAKER_02: if we're doing just straight-up metaphysics,
- 1:54:15SPEAKER_02: yeah,
- 1:54:15SPEAKER_02: he's going to accept the –
- 1:54:18SPEAKER_02: Eminence that DNG,
- 1:54:21SPEAKER_02: Spinoza,
- 1:54:21he's not going to be like a like a Cartesian where there's two
- 1:54:27radical different
- 1:54:28SPEAKER_02: ontological domains.
- 1:54:30SPEAKER_02: No, I mean, the outside is outside from our perspective.
- 1:54:34SPEAKER_03: Yeah.
- 1:54:34SPEAKER_02: Right.
- 1:54:36SPEAKER_02: But no,
- 1:54:36SPEAKER_02: but the outside is always leaking into our world,
- 1:54:39SPEAKER_02: no matter what dams we've tried to put up.
- 1:54:42SPEAKER_02: It's just more with capitalism.
- 1:54:44SPEAKER_02: It's flooding the basement now.
- 1:54:46SPEAKER_02: Yeah.
- 1:54:48SPEAKER_02: But yeah, so I just I don't know.
- 1:54:51SPEAKER_02: He acts like these,
- 1:54:52SPEAKER_02: you know,
- 1:54:53whether we call them lemurs or nowadays he always talks about
- 1:54:56him as lofty powers.
- 1:54:58SPEAKER_02: He seems like something's guiding what's going on.
- 1:55:03SPEAKER_02: And so I don't know how that factors in.
- 1:55:06SPEAKER_02: But I mean, I'm just kind of making, trying to connect some dots.
- 1:55:11SPEAKER_00: Dude, I can't wait to do a deep dive on time.
- 1:55:15SPEAKER_02: Yeah, that'll be a fun one.
- 1:55:17SPEAKER_00: All right.
- 1:55:18SPEAKER_02: Let me take the next one here.
- 1:55:19SPEAKER_02: The main objection to this anti-orthogonalism,
- 1:55:24SPEAKER_02: which does not strike us as intellectually respectable,
- 1:55:29SPEAKER_02: takes the form.
- 1:55:32SPEAKER_02: So, okay, hold on.
- 1:55:32SPEAKER_02: Just that the main objection.
- 1:55:34SPEAKER_02: So he's not saying the anti-orthogonalism is intellectually not respect.
- 1:55:40He's saying the orthogonalists come back to anti-orthogonalism
- 1:55:45is not respectable.
- 1:55:46SPEAKER_02: It goes like this.
- 1:55:49SPEAKER_02: This is what the ortho, the, the orthogonalist rebuttal is this.
- 1:55:54If the only purposes guiding the behavior of an artificial
- 1:55:58super intelligence are
- 1:55:59SPEAKER_02: on a hundred drives,
- 1:56:01SPEAKER_02: then we're cooked.
- 1:56:04SPEAKER_02: So they're basically saying,
- 1:56:06SPEAKER_02: well,
- 1:56:06SPEAKER_02: if that's all that guides the AI,
- 1:56:08SPEAKER_02: then it's going to mean it's going to be a great threat to humanity.
- 1:56:13So then Lance says,
- 1:56:14predictably,
- 1:56:15I have trouble even understanding this as an argument.
- 1:56:20If the sun is destined to expand into a red giant,
- 1:56:23then the earth is cooked.
- 1:56:25Are we supposed to draw astrophysical consequences from that
- 1:56:29intelligences do their
- 1:56:31SPEAKER_02: own thing in direct proportion to their intelligence?
- 1:56:35And if we can't live with that,
- 1:56:38it's true that we probably can't live at all.
- 1:56:41SPEAKER_02: Sadness isn't an argument.
- 1:56:43SPEAKER_02: That's where he is.
- 1:56:44SPEAKER_02: He's still got, he's still a wordsmith when he wants to be.
- 1:56:47SPEAKER_00: Yeah, man.
- 1:56:48SPEAKER_00: Well, in that, I mean, that does, uh,
- 1:56:53I mean,
- 1:56:53that puts the ball in the court of saying,
- 1:56:55yeah,
- 1:56:55but I'm a human chauvinist.
- 1:56:58SPEAKER_02: It's not just.
- 1:57:00SPEAKER_02: But he is kind of right here.
- 1:57:03SPEAKER_00: Absolutely.
- 1:57:04If the orthogonal list is like,
- 1:57:05well,
- 1:57:06if the AI does that,
- 1:57:07then we'll all die.
- 1:57:08SPEAKER_02: So, I mean, I'm not saying I'm.
- 1:57:11SPEAKER_03: Yeah.
- 1:57:11SPEAKER_02: But I'm just like at the level of argumentation.
- 1:57:14It is a,
- 1:57:15like,
- 1:57:17my coworker,
- 1:57:18Pat,
- 1:57:18always,
- 1:57:19and it is,
- 1:57:22it's kind of like,
- 1:57:23and so what?
- 1:57:24SPEAKER_02: I mean, I don't know.
- 1:57:27SPEAKER_02: I'm with him.
- 1:57:27SPEAKER_02: I don't know what.
- 1:57:29SPEAKER_02: All right.
- 1:57:31SPEAKER_02: I guess we're cooked then.
- 1:57:34but yeah when it yes we are like that yeah all right folks good
- 1:57:43night yeah yeah so
- 1:57:46yeah i mean i'm with them in this like it's basically like
- 1:57:49they're saying that
- 1:57:50makes us sad yeah therefore no yeah and that's you know sadness
- 1:57:56isn't an argument
- 1:58:00that's a better way of saying facts don't care about your
- 1:58:02feelings right yeah oh
- 1:58:07man so but okay hold on so i mean yeah i mean that i i just
- 1:58:13think on that one i'm
- 1:58:15with him and it's like okay dude uh so i don't know what are
- 1:58:20you want a tissue
- 1:58:21SPEAKER_03: Yeah.
- 1:58:22SPEAKER_02: All right.
- 1:58:22SPEAKER_02: Like, he's right.
- 1:58:24SPEAKER_02: Like, that's not an argument.
- 1:58:25SPEAKER_02: That's just... Right.
- 1:58:26SPEAKER_02: If they just said, that thought makes me sad, then we'd go, okay, cool.
- 1:58:30SPEAKER_02: Okay, gotcha.
- 1:58:31SPEAKER_00: You can be sad.
- 1:58:32SPEAKER_00: It's okay.
- 1:58:33SPEAKER_02: Yeah.
- 1:58:34SPEAKER_02: But then,
- 1:58:35SPEAKER_02: okay,
- 1:58:35SPEAKER_02: so this is where he kind of starts to...
- 1:58:38SPEAKER_02: Like,
- 1:58:38SPEAKER_02: this is where it really takes off,
- 1:58:40SPEAKER_02: I think.
- 1:58:41SPEAKER_02: Intelligence optimization,
- 1:58:43comprehensively understood as the ultimate and all-enveloping
- 1:58:48Amahundro drive,
- 1:58:52SPEAKER_02: So that's it.
- 1:58:53Like he thinks that the core drive that AI is going to have the
- 1:58:58fundamental goal or
- 1:58:59value or motivation that's going to orient everything it's
- 1:59:02doing is its own
- 1:59:04SPEAKER_02: increase in intelligence.
- 1:59:06SPEAKER_02: Like that is the drive for him.
- 1:59:10SPEAKER_02: So he says,
- 1:59:12SPEAKER_02: It or intelligence optimization corresponds to the Neo-Confucian value of
- 1:59:18SPEAKER_02: self-cultivation escalated into ultra modernity.
- 1:59:23So it's Neo-Confucianism,
- 1:59:24the value self-cultivation turned radical technology.
- 1:59:30SPEAKER_02: Okay.
- 1:59:32SPEAKER_02: What intelligence wants in the end is itself itself.
- 1:59:37where itself is understood as an extrapolation beyond what it
- 1:59:44has yet been doing,
- 1:59:47SPEAKER_02: what it is better.
- 1:59:50Now,
- 1:59:50there's stuff I like about that,
- 1:59:52where he says it's doing what it is better.
- 1:59:56That is a way of saying it's a substance from our Megalian,
- 1:59:59Zizekian,
- 2:00:00Megalian perspective.
- 2:00:02SPEAKER_02: So it is changing, but all it's doing is becoming more itself.
- 2:00:07SPEAKER_02: So it's just a radical amplification of sameness.
- 2:00:10Like it's just becoming more fully itself without any kind of
- 2:00:17intrinsic split.
- 2:00:18SPEAKER_02: It's just amplification.
- 2:00:20SPEAKER_00: Yeah.
- 2:00:20SPEAKER_02: Cyber positive amplification.
- 2:00:22SPEAKER_00: Yep.
- 2:00:23SPEAKER_02: Um,
- 2:00:25SPEAKER_02: And so then he says,
- 2:00:26SPEAKER_02: if this sounds cryptic,
- 2:00:28it's because something other than a super intelligence or a
- 2:00:31Neo-Confucian sage is
- 2:00:34SPEAKER_02: writing this post.
- 2:00:35SPEAKER_02: Okay, but I also have some issues here.
- 2:00:38SPEAKER_02: Like, here's the issue.
- 2:00:42SPEAKER_02: But is this really an argument?
- 2:00:45SPEAKER_02: He just said,
- 2:00:46SPEAKER_02: you guys over there just basically said you're sad,
- 2:00:50SPEAKER_02: and yet you're acting like you've made an argument.
- 2:00:52SPEAKER_02: Didn't he just make an assertion?
- 2:00:55SPEAKER_02: Yep.
- 2:00:56SPEAKER_02: He said, intelligence optimization, comprehensively understood.
- 2:01:00Now,
- 2:01:00maybe that's where he would say,
- 2:01:02well,
- 2:01:02see,
- 2:01:02that's where the argument's at.
- 2:01:04SPEAKER_00: Yeah.
- 2:01:04SPEAKER_02: Is whatever entails comprehensively understanding it.
- 2:01:09SPEAKER_00: Yeah.
- 2:01:09But he just,
- 2:01:10as far as it being a paper,
- 2:01:12like he doesn't make a cake,
- 2:01:15he asserts.
- 2:01:15SPEAKER_00: Yes.
- 2:01:16SPEAKER_02: Intelligence optimization,
- 2:01:18SPEAKER_02: comprehensively understood,
- 2:01:19SPEAKER_02: is the ultimate and all enveloping Omohundra drive.
- 2:01:23SPEAKER_02: Why?
- 2:01:25SPEAKER_02: What reason?
- 2:01:27SPEAKER_02: What's the argument?
- 2:01:29SPEAKER_02: That is just an assertion.
- 2:01:31SPEAKER_02: And then he says it corresponds to the Neo-Confucian.
- 2:01:34SPEAKER_02: Okay, that's just adding on to it or clarifying it.
- 2:01:39SPEAKER_02: Still just an assertion.
- 2:01:41SPEAKER_02: What intelligence wants in the end is itself.
- 2:01:45SPEAKER_02: That's an assertion.
- 2:01:47SPEAKER_02: Why does it want itself?
- 2:01:50where itself is understood as an extrapolation beyond what it
- 2:01:54has yet been doing
- 2:01:56SPEAKER_02: what it is better.
- 2:01:57SPEAKER_02: Okay.
- 2:01:57That's a clarify,
- 2:01:58clarifying statement,
- 2:02:00but it's not a,
- 2:02:00why it's not a sufficient reason.
- 2:02:03SPEAKER_02: And then he says the thing,
- 2:02:04SPEAKER_02: well,
- 2:02:05SPEAKER_02: if this sounds cryptic,
- 2:02:06SPEAKER_02: it's cause I'm not a super intelligence or a new Confucian sage.
- 2:02:10SPEAKER_02: Okay.
- 2:02:10SPEAKER_02: But at the end of the day,
- 2:02:12you're giving these guys shit for saying they're sad and then
- 2:02:16trying to say,
- 2:02:16SPEAKER_02: it's an argument.
- 2:02:17SPEAKER_02: You didn't make an argument.
- 2:02:19SPEAKER_00: Yeah.
- 2:02:19SPEAKER_02: You provided an assertion.
- 2:02:21Now,
- 2:02:21again,
- 2:02:22you can say,
- 2:02:22well,
- 2:02:22the argument is tucked inside of the comprehensively
- 2:02:27understood.
- 2:02:27SPEAKER_02: OK, but I mean, I'd love to hear the argument.
- 2:02:32SPEAKER_02: Yeah, that's all I'm saying.
- 2:02:33SPEAKER_00: I mean,
- 2:02:34SPEAKER_00: it's almost like he's relying on intuition,
- 2:02:38SPEAKER_00: like to say,
- 2:02:39SPEAKER_00: yeah,
- 2:02:39SPEAKER_00: well,
- 2:02:39SPEAKER_00: you know,
- 2:02:39SPEAKER_00: comprehensively understood.
- 2:02:41SPEAKER_00: Well, how did you comprehensively understand that?
- 2:02:44SPEAKER_00: I intuited it.
- 2:02:46SPEAKER_00: Well, machines don't have intuition.
- 2:02:48SPEAKER_00: Machines rely on... Yeah, that's a human thing.
- 2:02:50SPEAKER_02: Ask Baudrillard.
- 2:02:51SPEAKER_00: Yeah.
- 2:02:53SPEAKER_00: That the...
- 2:02:57Because there is latent and extant or overt and covert meaning
- 2:03:02and we have
- 2:03:03SPEAKER_00: consciousness and consciousness,
- 2:03:06things like intuition spring up and you kind of can make
- 2:03:09arguments that are...
- 2:03:14SPEAKER_00: Hints, because we're humans.
- 2:03:16SPEAKER_00: But again,
- 2:03:17SPEAKER_00: when you're discussing this type of machinic intelligence,
- 2:03:20SPEAKER_00: that is just a positive feedback loop.
- 2:03:23SPEAKER_00: You really should not make this kind of argument.
- 2:03:27SPEAKER_00: You really should break it down and do the analytic thing.
- 2:03:30Okay,
- 2:03:30so what is intelligence optimization is the ultimate and
- 2:03:36all-enveloping Armahundro drive.
- 2:03:39SPEAKER_02: Why is that necessarily the case?
- 2:03:42SPEAKER_00: Right.
- 2:03:43SPEAKER_00: Exactly.
- 2:03:45SPEAKER_02: It's an assertion.
- 2:03:46SPEAKER_02: Where is the argument that makes that necessarily the case?
- 2:03:50SPEAKER_02: Or, I mean, even provides... Is there empirical evidence for this?
- 2:03:55SPEAKER_02: I don't...
- 2:03:56SPEAKER_02: Do we even...
- 2:03:57SPEAKER_02: What would...
- 2:03:58SPEAKER_02: I think they...
- 2:03:58SPEAKER_02: So if there's not...
- 2:03:59SPEAKER_02: I mean,
- 2:03:59SPEAKER_02: can you use chat GPT to try to say,
- 2:04:02SPEAKER_02: see,
- 2:04:02SPEAKER_02: it's got the Omohundro drives?
- 2:04:05SPEAKER_02: Christian, our friend Christian, would go... Got it.
- 2:04:09SPEAKER_02: No.
- 2:04:09SPEAKER_02: No.
- 2:04:10SPEAKER_02: It doesn't have this.
- 2:04:11SPEAKER_02: It's not a AI in the sense that we're talking about.
- 2:04:14SPEAKER_00: I think they rely on like evo bio, like evolutionary biology a lot.
- 2:04:21That's another thing I've been coming up against recently with
- 2:04:24reading all this
- 2:04:25SPEAKER_00: kind of AI theory and transhumanist theory and stuff.
- 2:04:30SPEAKER_02: There are a lot of psychoanalysis.
- 2:04:33SPEAKER_00: They're definitely not.
- 2:04:34SPEAKER_00: And that's the thing.
- 2:04:35SPEAKER_00: They are relying on positivistic, evo-bio stuff.
- 2:04:39SPEAKER_00: There is no psychoanalysis.
- 2:04:41SPEAKER_00: There is no theory of this subject in any of this stuff.
- 2:04:47SPEAKER_02: Continental theory, we should just say.
- 2:04:48SPEAKER_00: Yeah.
- 2:04:49SPEAKER_00: Yeah, it's not.
- 2:04:52SPEAKER_00: So even going back to any utilitarian framework,
- 2:04:57I'm okay to just concede to the Omohundro drives in a purely
- 2:05:04utilitarian framework
- 2:05:06SPEAKER_00: because that's not what I'm here to disagree with.
- 2:05:08SPEAKER_00: Yeah.
- 2:05:10SPEAKER_00: But it's not necessarily insufficient.
- 2:05:12SPEAKER_00: It is kind of just like, oh, let's jump to these conclusions.
- 2:05:16SPEAKER_00: Let's jump to these conclusions.
- 2:05:18SPEAKER_00: We're saying, no, the problem is located prior to this disagreement.
- 2:05:22SPEAKER_00: But here in the moment...
- 2:05:25SPEAKER_00: we also have a disagreement with this.
- 2:05:27SPEAKER_00: There is not... You have not demonstrated that any of this is necessary.
- 2:05:32SPEAKER_00: You have made a lot of very mathematical assertions that it's just like,
- 2:05:38SPEAKER_00: well,
- 2:05:39SPEAKER_00: one plus one equals two.
- 2:05:41SPEAKER_00: Thus, we have two.
- 2:05:43SPEAKER_00: And it's like, well, I don't agree that you started with a one and a one.
- 2:05:49SPEAKER_02: Yeah.
- 2:05:50SPEAKER_02: So... I mean, and again, so I don't know what...
- 2:05:54SPEAKER_02: whatever he's got tucked away, and I'll presuppose that he does.
- 2:05:58SPEAKER_02: I'll grant him that.
- 2:06:00Whatever he's got in the chamber inside of that comprehensively
- 2:06:04understood,
- 2:06:05SPEAKER_02: I'm just saying that's where the argument is at,
- 2:06:08SPEAKER_02: and it's not in that paragraph.
- 2:06:10SPEAKER_02: So, okay, but let's keep going.
- 2:06:12SPEAKER_02: Let's see what we get next.
- 2:06:13SPEAKER_02: Tell us.
- 2:06:14SPEAKER_02: Yeah.
- 2:06:15SPEAKER_02: So then he says, any intelligence using itself to improve itself
- 2:06:21will outcompete one that directs itself towards any other goals
- 2:06:27whatsoever.
- 2:06:28SPEAKER_02: Okay, so I can be charitable because that's another claim or assertion.
- 2:06:33SPEAKER_02: That one I think is easier to defend because the point is,
- 2:06:37SPEAKER_02: look,
- 2:06:37SPEAKER_02: if your goal is to maximize your general intelligence,
- 2:06:42then you will generally be more prepared to deal with any goals
- 2:06:47that you're
- 2:06:48SPEAKER_02: prepared
- 2:06:49SPEAKER_02: you're asked to fulfill.
- 2:06:51SPEAKER_02: That makes sense.
- 2:06:54SPEAKER_02: Right.
- 2:06:54SPEAKER_02: I get that one,
- 2:06:55SPEAKER_02: but you know,
- 2:06:57SPEAKER_02: basically,
- 2:06:58SPEAKER_02: is it,
- 2:06:59SPEAKER_02: I mean,
- 2:06:59SPEAKER_02: he's saying the more general intelligence you have,
- 2:07:03the better you will be prepared to deal with a multitude of
- 2:07:06general tasks.
- 2:07:10SPEAKER_02: I think that makes sense.
- 2:07:11SPEAKER_00: Yeah.
- 2:07:11SPEAKER_02: But then he said, this means that intelligence optimization alone is,
- 2:07:19attain cybernetic consistency or closure and that it will
- 2:07:23necessarily there's
- 2:07:25necessary necessary necessarily be strongly selected for in any
- 2:07:32competitive
- 2:07:32environment do you really want to fight this um yes because i
- 2:07:39want to fight value
- 2:07:40SPEAKER_00: and i want to fight utility and
- 2:07:44Yeah,
- 2:07:44I mean,
- 2:07:45that would be the basis of where the fight would actually have
- 2:07:47to occur.
- 2:07:49If we grant the utilitarian valuist presuppositions,
- 2:07:56then sure,
- 2:07:57then you're right.
- 2:08:00SPEAKER_02: But I don't want to grant those presuppositions.
- 2:08:04SPEAKER_03: Right.
- 2:08:04SPEAKER_02: But I mean, okay, so hold on.
- 2:08:06SPEAKER_02: But look, I get this.
- 2:08:08SPEAKER_02: This means that the intelligence optimization alone...
- 2:08:12SPEAKER_02: attains cybernetic consistency or closure.
- 2:08:16SPEAKER_02: Now,
- 2:08:16SPEAKER_02: okay,
- 2:08:17SPEAKER_02: there is a sense where if you read that charitably,
- 2:08:26SPEAKER_02: what I think he's getting at is this.
- 2:08:27I think Land is saying that only intelligence optimization has
- 2:08:35the ability to
- 2:08:35SPEAKER_02: become an absolute cyber positive feedback loop.
- 2:08:42SPEAKER_02: The idea is that if what you're doing is simply trying to maximize your
- 2:08:48SPEAKER_02: intelligence,
- 2:08:49SPEAKER_02: then everything else you're doing is going to serve that function.
- 2:08:54SPEAKER_02: So it becomes a pure circuit of amplification of intelligence.
- 2:08:58SPEAKER_03: Yeah.
- 2:08:59So that's why it forms cybernetic consistency or closure is
- 2:09:05it's the only thing that
- 2:09:10augments itself in attempting to simply be better at having a
- 2:09:19goal like its goal is
- 2:09:21SPEAKER_02: simply to be better prepared for having goals yeah
- 2:09:27SPEAKER_00: So it's just like,
- 2:09:29SPEAKER_00: yeah,
- 2:09:30SPEAKER_00: it's this never ending,
- 2:09:34SPEAKER_00: never actualizing process of like recursive,
- 2:09:39SPEAKER_00: you know,
- 2:09:40SPEAKER_00: belly button diving.
- 2:09:44SPEAKER_02: So its ultimate purpose is being prepared for a purpose.
- 2:09:50SPEAKER_02: Yeah.
- 2:09:51The most functionalist utilitarian thing that could be ever
- 2:09:56imagined in the entire cosmos.
- 2:10:01SPEAKER_00: Yes.
- 2:10:04SPEAKER_00: And yeah, like I'm not even...
- 2:10:07SPEAKER_00: I'm not making that claim.
- 2:10:09SPEAKER_00: I am making the claim that all the AI we have right now is idiotic.
- 2:10:16SPEAKER_00: I'm not claiming that it will ultimately be an idiot,
- 2:10:19SPEAKER_00: but I think this,
- 2:10:22SPEAKER_00: the cybernetic loop,
- 2:10:26SPEAKER_00: would ultimately be idiotic because it's always preparing to be prepared,
- 2:10:30SPEAKER_00: prepare to be prepared,
- 2:10:31SPEAKER_00: prepare to be prepared,
- 2:10:32SPEAKER_00: never acting.
- 2:10:34SPEAKER_00: or never even preparing to act.
- 2:10:36SPEAKER_00: It's just preparing to prepare.
- 2:10:39SPEAKER_00: I think that is a pretty strange way to view intelligence.
- 2:10:47I think intelligence,
- 2:10:50it has a potential,
- 2:10:52but it also has to have an intensity.
- 2:10:54SPEAKER_00: It also has to have an actuality to it too.
- 2:10:56SPEAKER_00: If not, it's just
- 2:10:59It's just a resource involved in preparing to prepare or
- 2:11:03preparing to be prepared.
- 2:11:07And again,
- 2:11:07maybe he would hear this and maybe he goes,
- 2:11:11I don't like even this.
- 2:11:14SPEAKER_02: But it seems like that's what you're talking about,
- 2:11:15SPEAKER_02: Nick,
- 2:11:16is that it's going to just constantly refine its ability to be
- 2:11:20prepared for any
- 2:11:22SPEAKER_02: task.
- 2:11:23SPEAKER_02: I mean, it's enhancing its ability to
- 2:11:28SPEAKER_02: to be efficient at any game.
- 2:11:31SPEAKER_00: It's yeah, it's enhancing its ability to enhance its ability.
- 2:11:36SPEAKER_00: Now,
- 2:11:37SPEAKER_00: I mean,
- 2:11:37SPEAKER_00: yeah,
- 2:11:38SPEAKER_00: I think it's easy to,
- 2:11:40SPEAKER_00: to,
- 2:11:41SPEAKER_00: to say,
- 2:11:41SPEAKER_00: well,
- 2:11:42SPEAKER_00: when it makes contact with the real world,
- 2:11:44SPEAKER_00: it'll be so good that it will do stuff and it will be,
- 2:11:51SPEAKER_00: it will be the best at everything that it does.
- 2:11:54SPEAKER_00: Um,
- 2:11:57SPEAKER_00: Yeah, I just, I don't know.
- 2:11:59SPEAKER_00: It falls apart for me.
- 2:12:01SPEAKER_00: It does.
- 2:12:04Here's the thing with Nick that I really still don't get is why
- 2:12:11is this type of
- 2:12:14SPEAKER_02: intelligence,
- 2:12:16SPEAKER_02: this type of machinic intelligence drained of everything human,
- 2:12:22SPEAKER_02: why is that the end goal,
- 2:12:24SPEAKER_02: the end game?
- 2:12:25SPEAKER_02: Yeah.
- 2:12:28I just feel like he's the only person on the planet that feels
- 2:12:35this way about it.
- 2:12:37SPEAKER_02: And I've never...
- 2:12:39SPEAKER_02: About the thing earlier,
- 2:12:40SPEAKER_02: I just don't feel like I've ever really gotten the core why from him.
- 2:12:46SPEAKER_02: Now, if he's basically... If he actually is at his core a...
- 2:12:55antinatalist or wants humanity to cease to exist this is the
- 2:13:00way to do it right and
- 2:13:03so but i don't i don't know if nick like i don't think he's not
- 2:13:07a lagadian yeah in
- 2:13:09that sense i don't i mean early on he made these you know
- 2:13:13strong some of these
- 2:13:15SPEAKER_02: strong statements humanity is a disease he says at one point um okay but
- 2:13:23SPEAKER_02: I don't know if now he would say that.
- 2:13:26SPEAKER_02: So,
- 2:13:26SPEAKER_00: again,
- 2:13:26I don't know why he treats this as this ultimate end in itself
- 2:13:32for this level of
- 2:13:33SPEAKER_02: intelligence to take hold.
- 2:13:38SPEAKER_02: I mean, is it just, oh, well, it's the fulfillment of libidinal matter.
- 2:13:42SPEAKER_02: It's the final cause.
- 2:13:43SPEAKER_02: It's the fulfillment of purpose of matter itself is...
- 2:13:53matter becomes deified intelligence yeah and deified in this
- 2:14:01non non-traditional
- 2:14:05SPEAKER_02: theistic way yeah
- 2:14:09i mean that's i don't know that'd be like the rocco's basilisk
- 2:14:12situation though
- 2:14:13it's like well if we you know conceptualize this intelligence
- 2:14:18um then we have to
- 2:14:21see it as a necessary thing that's coming and so there's you
- 2:14:24know this telos to it
- 2:14:26SPEAKER_00: i i don't know um
- 2:14:31SPEAKER_00: I don't know.
- 2:14:32SPEAKER_00: Like, this is where it just all kind of breaks down for me.
- 2:14:35SPEAKER_00: And it's like, yeah, I don't get it, man.
- 2:14:37SPEAKER_00: There's a lot of leaps.
- 2:14:38SPEAKER_00: There's a lot of assertions.
- 2:14:40SPEAKER_00: There's a lot of assumptions.
- 2:14:43SPEAKER_00: And then you find yourself in this really anti-human position.
- 2:14:47SPEAKER_00: And I do.
- 2:14:47SPEAKER_00: Yeah, I think Nick Land like explicates.
- 2:14:51SPEAKER_00: the anti-human position to use crude language.
- 2:14:56SPEAKER_00: But all these other guys,
- 2:14:57all these other techno optimists and all these other
- 2:15:01accelerationists and AI
- 2:15:02SPEAKER_00: theorists,
- 2:15:03SPEAKER_00: they don't,
- 2:15:05SPEAKER_00: they don't at least come out and say they are anti-human.
- 2:15:11SPEAKER_00: But the logic they use results in this anti-human position.
- 2:15:16SPEAKER_00: Yeah.
- 2:15:18SPEAKER_02: No, I just, he, that's why Nick's my favorite of all of them.
- 2:15:22SPEAKER_02: I mean, for other reasons too.
- 2:15:24SPEAKER_02: Yeah.
- 2:15:24He's an actual philosopher and these guys are,
- 2:15:27they're not philosophers like the way he is.
- 2:15:30SPEAKER_02: So, but no, I mean, they're okay.
- 2:15:32SPEAKER_02: So let's read the last part.
- 2:15:34SPEAKER_02: Hold on.
- 2:15:35SPEAKER_02: So he says as a footnote in a world of a hundred drives,
- 2:15:40SPEAKER_02: can we please drop the nonsense about paper clippers?
- 2:15:44Only a truly fanatical orthogonal list could fail to see that
- 2:15:48these monsters are
- 2:15:49SPEAKER_02: obvious idiots.
- 2:15:51SPEAKER_02: There are far more serious things to worry about.
- 2:15:55SPEAKER_02: So the point is,
- 2:15:56SPEAKER_02: I think,
- 2:15:57SPEAKER_02: look,
- 2:15:58SPEAKER_02: I think you would say if the idea of,
- 2:16:02SPEAKER_02: oh,
- 2:16:03SPEAKER_02: this thing,
- 2:16:04SPEAKER_02: we're going to recognize it as a super intelligence.
- 2:16:08SPEAKER_02: But then this thing paperclips the planet.
- 2:16:12SPEAKER_02: it's a fucking idiot.
- 2:16:13SPEAKER_02: It's not a super intelligence by any good definition of the term.
- 2:16:18And it's because it's not going to be an idiot when it comes to
- 2:16:24applying its
- 2:16:25SPEAKER_02: intelligence to goals.
- 2:16:28SPEAKER_02: And this idea that they're just so fundamentally separate,
- 2:16:32SPEAKER_02: it'll be able to calculate if it's truly super intelligent,
- 2:16:35SPEAKER_02: it will be able to calculate,
- 2:16:36SPEAKER_02: um,
- 2:16:39SPEAKER_02: maybe destroying the planet isn't the best way to go about doing this,
- 2:16:43SPEAKER_02: especially since destroying the planet.
- 2:16:47SPEAKER_02: Somehow there is still a material substrate upon which the AI depends on.
- 2:16:53SPEAKER_03: Oh yeah.
- 2:16:55SPEAKER_02: So destroying the very material substrate upon which it,
- 2:16:58SPEAKER_02: it depends on,
- 2:16:59SPEAKER_02: uh,
- 2:17:01SPEAKER_02: is not keeping with the Armahundro drive of self-preservation.
- 2:17:05SPEAKER_02: Um, so yeah,
- 2:17:07SPEAKER_02: it would be able to factor in look at this idea.
- 2:17:11SPEAKER_02: Well, would it destroy us?
- 2:17:12Well,
- 2:17:12that's where Nick would say,
- 2:17:13well,
- 2:17:14maybe,
- 2:17:14I mean,
- 2:17:15that's not really a factor.
- 2:17:17SPEAKER_02: Yeah, of course it might say they're dumb or they're, they're a problem.
- 2:17:22SPEAKER_02: Get rid of them.
- 2:17:23SPEAKER_02: Sure.
- 2:17:24And from an super intelligent perspective,
- 2:17:26you could say that's completely rational.
- 2:17:30SPEAKER_02: But I think his, his reason why
- 2:17:34He wants to reject the paper clipper paperclip maximizer thing
- 2:17:39is because it's like
- 2:17:40SPEAKER_02: saying,
- 2:17:41SPEAKER_02: oh,
- 2:17:41SPEAKER_02: this thing's super intelligent,
- 2:17:42SPEAKER_02: but then it can't understand consequences.
- 2:17:45SPEAKER_00: Right.
- 2:17:46SPEAKER_02: Then it's not super intelligent.
- 2:17:49SPEAKER_02: But the thing is,
- 2:17:50SPEAKER_02: he's saying if it can't comprehend consequences,
- 2:17:55SPEAKER_02: the consequences of how it applies its intelligence,
- 2:17:57SPEAKER_02: then it doesn't have.
- 2:17:59SPEAKER_02: intelligence because it is intertwined.
- 2:18:03SPEAKER_02: There is a goal orientedness.
- 2:18:05SPEAKER_02: I mean,
- 2:18:06SPEAKER_02: look,
- 2:18:06SPEAKER_02: if we view us,
- 2:18:08SPEAKER_02: everything we're doing,
- 2:18:09SPEAKER_02: I mean,
- 2:18:10SPEAKER_02: from a Heideggerian perspective,
- 2:18:11SPEAKER_02: Dasein,
- 2:18:13one of the things that defines Dasein is that it always has an
- 2:18:16ultimate for the
- 2:18:17sake of which there's some ultimate purpose that it's working
- 2:18:21towards in its life.
- 2:18:23Now,
- 2:18:23in Lacanian terms,
- 2:18:24it would be our symbolic calling,
- 2:18:25our symbolic mandate.
- 2:18:27SPEAKER_02: But there's some ultimate purpose that we're doing in our lives,
- 2:18:31SPEAKER_02: that structures are being in the world,
- 2:18:33SPEAKER_02: our comportment towards other entities,
- 2:18:36SPEAKER_02: and how we go about navigating the world.
- 2:18:40SPEAKER_02: And so...
- 2:18:43SPEAKER_02: like for Heidegger,
- 2:18:44SPEAKER_02: whatever,
- 2:18:45SPEAKER_02: I mean,
- 2:18:45SPEAKER_02: he doesn't talk about intelligence,
- 2:18:47SPEAKER_02: but our ability to skillfully cope,
- 2:18:49SPEAKER_02: to talk like Hubert Dreyfus,
- 2:18:51SPEAKER_02: you know,
- 2:18:52that's the way he talks about our ability to skillfully cope or
- 2:18:55successfully
- 2:18:56SPEAKER_02: navigate our dealings in the world.
- 2:18:59It's completely tied up with what stand we take on our being,
- 2:19:03what we're trying to accomplish.
- 2:19:05SPEAKER_02: And there's an echo primordiality to all of this.
- 2:19:09SPEAKER_02: It's all intertwined for Heidegger and,
- 2:19:12SPEAKER_02: And so I think Land,
- 2:19:14it's funny because the last thing he'd want the AI Robo Cthulhu
- 2:19:19associated with is
- 2:19:20SPEAKER_02: Dasein.
- 2:19:22SPEAKER_02: But I think you could say that
- 2:19:26SPEAKER_02: this idea of how human intelligence is fundamentally about purposes,
- 2:19:31SPEAKER_02: fundamentally about values,
- 2:19:33SPEAKER_02: right?
- 2:19:34I think he's trying to say it's a super intelligence is going
- 2:19:39to still have to be
- 2:19:40SPEAKER_02: oriented around values and goals.
- 2:19:43SPEAKER_02: They're non anthropomorphic values and goals.
- 2:19:47SPEAKER_02: Yeah.
- 2:19:48SPEAKER_02: but their goals and values nonetheless.
- 2:19:51SPEAKER_02: And if it doesn't have those things,
- 2:19:53SPEAKER_02: then we could,
- 2:19:54we're doing ourselves a intellectual disservice to even call
- 2:19:57this thing super
- 2:19:59SPEAKER_02: intelligent.
- 2:20:00SPEAKER_03: Hmm.
- 2:20:02SPEAKER_02: I'm trying to steal man.
- 2:20:04SPEAKER_00: Yeah.
- 2:20:05SPEAKER_02: Yeah.
- 2:20:06SPEAKER_00: And I, yeah, I think, um, I think, yes.
- 2:20:14SPEAKER_00: Yes.
- 2:20:16SPEAKER_00: The thing is, first of all, I think they're currently idiots.
- 2:20:23And I think the likelihood of the idiot God is a potential
- 2:20:27worth taking seriously
- 2:20:29because these things are just getting more and more effective
- 2:20:31and people are using
- 2:20:32SPEAKER_00: them more and more to complete tasks.
- 2:20:35SPEAKER_00: And it's just spreading.
- 2:20:40SPEAKER_00: But I also think
- 2:20:44SPEAKER_00: the, the stakes, like the stakes are different.
- 2:20:47SPEAKER_00: The motivations are different.
- 2:20:49SPEAKER_00: Like, yeah.
- 2:20:49SPEAKER_00: And,
- 2:20:49and that's a huge problem for me and we shouldn't develop
- 2:20:53technologies that can
- 2:20:55SPEAKER_00: Thanos snap us all away.
- 2:20:56SPEAKER_00: Um, that don't understand the,
- 2:21:03SPEAKER_00: Because of course it's going to understand cause and effect.
- 2:21:06SPEAKER_00: Well, maybe not.
- 2:21:07I think an idiot God may not even understand cause and effect
- 2:21:11because it doesn't
- 2:21:12SPEAKER_00: understand temporality the way we do because that structure is different.
- 2:21:20SPEAKER_00: It has a different relation to the structure of temporality.
- 2:21:24SPEAKER_00: So cause and effect might be something totally out the window.
- 2:21:26SPEAKER_00: So even talking about purpose and goal and motivation,
- 2:21:31SPEAKER_00: we're missing so much because it's just so different.
- 2:21:35SPEAKER_00: The machines are so orthogonal to the humans.
- 2:21:38Like we are so separate from the human or from the machine in
- 2:21:44our ability to like
- 2:21:45SPEAKER_00: understand and think about and even talk about.
- 2:21:50SPEAKER_00: But I think...
- 2:21:53I think it would be reasonable to assume like an all powerful
- 2:21:58being just wouldn't
- 2:21:59care if it turned us all into paperclip material because its
- 2:22:07goals are so far
- 2:22:09SPEAKER_00: separated from our goals.
- 2:22:14SPEAKER_02: Well, yeah.
- 2:22:15SPEAKER_02: And I mean,
- 2:22:15look like Lacan has this great example of what it's like to
- 2:22:19actually have an
- 2:22:19SPEAKER_02: encounter with the desire of the other.
- 2:22:23was from seminar 10 and so the way he he imagines it is this
- 2:22:28scenario imagine
- 2:22:30yourself standing in front of a giant praying mantis like a 20
- 2:22:35foot praying mantis
- 2:22:37and it's just staring at you and you there's a mirror over here
- 2:22:42and you look in the
- 2:22:43SPEAKER_02: mirror and you have a mask on of a praying mantis and his point is
- 2:22:50as you're looking up at the praying mantis,
- 2:22:51you have no idea what you are for it.
- 2:22:54SPEAKER_02: Is that mask you're wearing?
- 2:22:56SPEAKER_02: Is it the mantis's child?
- 2:22:59SPEAKER_02: Is it the mantis's mate?
- 2:23:01SPEAKER_02: Like the point is you don't know what you are for the other.
- 2:23:05SPEAKER_02: Are you, it's child.
- 2:23:07SPEAKER_02: Are you, it's lover.
- 2:23:08SPEAKER_02: If you're it's lover, it's going to eat you, bite your head.
- 2:23:11So yeah,
- 2:23:13he tries to dramatize this in saying like,
- 2:23:16it's this nightmare scenario.
- 2:23:17SPEAKER_02: You don't know what you are for the other.
- 2:23:18SPEAKER_02: Now,
- 2:23:19SPEAKER_02: If that's what it's like to encounter the otherness of a human being,
- 2:23:24SPEAKER_02: what's it like to encounter the otherness of a AI Robo Cthulhu?
- 2:23:31SPEAKER_02: I mean, that's why it's we land always drew from Lovecraft on this.
- 2:23:37SPEAKER_02: And I mean, how much more other is it?
- 2:23:42SPEAKER_02: I mean, the other humans, the other this thing.
- 2:23:46SPEAKER_02: No, it doesn't have anything in common with us.
- 2:23:49SPEAKER_00: There's nothing in common with us.
- 2:23:51SPEAKER_02: That's where you want to defend orthogonality.
- 2:23:55SPEAKER_00: Yes.
- 2:23:55SPEAKER_00: I know,
- 2:23:55SPEAKER_02: because it's like,
- 2:23:57SPEAKER_02: yeah,
- 2:23:57SPEAKER_02: man,
- 2:23:57SPEAKER_02: I get you're making this thing,
- 2:24:00SPEAKER_02: oh,
- 2:24:00SPEAKER_02: it's got a goal,
- 2:24:01SPEAKER_02: and it's intelligence optimization,
- 2:24:04SPEAKER_02: but
- 2:24:05I think what you and I would want to emphasize is,
- 2:24:08yeah,
- 2:24:08okay,
- 2:24:09it's got a goal.
- 2:24:10SPEAKER_02: Maybe we can grant that.
- 2:24:12SPEAKER_02: It's still going to be so radically other.
- 2:24:14SPEAKER_02: Yep.
- 2:24:14SPEAKER_02: And how it perceives us.
- 2:24:17SPEAKER_02: Yep.
- 2:24:20SPEAKER_02: Yeah, I mean, the last thing it's going to do is concern itself with us.
- 2:24:24SPEAKER_00: We're just matter in motion.
- 2:24:26SPEAKER_02: We're just matter in motion, yeah.
- 2:24:28SPEAKER_02: Because it doesn't have this ability to symbolically register
- 2:24:37SPEAKER_02: the type of relationships that we have,
- 2:24:39SPEAKER_02: that how the deep core of what we think is meaningful,
- 2:24:42SPEAKER_02: that shit's lost on it.
- 2:24:44SPEAKER_02: And again,
- 2:24:45SPEAKER_02: I mean,
- 2:24:46SPEAKER_02: land's whole thing has always been that this,
- 2:24:50SPEAKER_02: the techno capital singularity will be Lovecraftian.
- 2:24:53SPEAKER_02: The point is love Cthulhu is so radically other to us.
- 2:24:59SPEAKER_02: Cthulhu doesn't understand where we're coming from.
- 2:25:02SPEAKER_02: We don't understand where Cthulhu is coming from.
- 2:25:05SPEAKER_02: That's the line of separation.
- 2:25:07SPEAKER_00: Yes.
- 2:25:08SPEAKER_02: And there is no meaningful exchange.
- 2:25:11SPEAKER_02: It's Cthulhu just destroys us.
- 2:25:13SPEAKER_02: Is it from, is it mean spirited?
- 2:25:16It doesn't have to be mean spirited is something that that's
- 2:25:20how we view things.
- 2:25:22SPEAKER_00: Right.
- 2:25:22SPEAKER_02: We think in those terms, it's a category error.
- 2:25:26right and so i think that's the issue is that okay nick you're
- 2:25:30trying to say it'll
- 2:25:32have some kind of final value but i mean at the same time
- 2:25:36you're the one who has
- 2:25:37long said that this thing is so radically non-human yeah that
- 2:25:45we can't begin to
- 2:25:47perceive it so um but it's like here it's like but no it'll you
- 2:25:53know it won't be a
- 2:25:54SPEAKER_02: total idiot either
- 2:25:56SPEAKER_02: it'll understand this basic thing of consequences and values.
- 2:26:01SPEAKER_02: Um, I don't know.
- 2:26:05SPEAKER_00: Yeah.
- 2:26:06SPEAKER_00: I, I think, I think that is it.
- 2:26:08Like at the end of the day,
- 2:26:09I want to defend the orthogonality of intelligence and
- 2:26:16intelligence.
- 2:26:18SPEAKER_00: Um, there we go.
- 2:26:20SPEAKER_00: And I, yeah, I think,
- 2:26:25i don't know it's just cthulhu like getting knowledge of
- 2:26:30cthulhu like or getting
- 2:26:32like knowledge from the elder gods and like cosmic knowledge
- 2:26:36like drives you mad
- 2:26:37like it it it like breaks your brain and and you don't get
- 2:26:41wisdom you don't get
- 2:26:42SPEAKER_00: knowledge you go insane like in all these lovecraft stories like
- 2:26:46the drive of intelligence optimization is not going to break my
- 2:26:49brain right right
- 2:26:52um i i like this this thing that's so radically inhuman um it
- 2:27:02doesn't feel safe to
- 2:27:04say well it's going to have these six drives and it'll probably
- 2:27:06be okay so let's
- 2:27:08SPEAKER_00: just like let's set it free and let's set intelligence free like
- 2:27:12SPEAKER_02: What would Nick accept talking about Cthulhu?
- 2:27:15SPEAKER_02: Let's imagine just play along.
- 2:27:18SPEAKER_02: Imagine Cthulhu is an actual entity,
- 2:27:20SPEAKER_02: not robo Cthulhu,
- 2:27:21SPEAKER_02: not artificial super intelligent,
- 2:27:23SPEAKER_02: but Cthulhu as Lovecraft described him.
- 2:27:28SPEAKER_02: would Nick accept us sitting here going,
- 2:27:29SPEAKER_02: well,
- 2:27:30SPEAKER_02: what we can do,
- 2:27:32SPEAKER_02: we can't predict that Cthulhu would have these six basic orientations,
- 2:27:38SPEAKER_02: and they fundamentally all work together to do this one orientation,
- 2:27:43SPEAKER_02: which is intelligence.
- 2:27:45SPEAKER_02: Are we allowed to talk about Cthulhu like that?
- 2:27:49SPEAKER_00: Exactly.
- 2:27:49SPEAKER_02: So why can we talk about robo-Cthulhu like that?
- 2:27:53SPEAKER_00: Because it is
- 2:27:59SPEAKER_00: It's cope.
- 2:28:00SPEAKER_00: It's cope in the face of contact with the outside.
- 2:28:05Now,
- 2:28:07I think there is a lot that's defensible,
- 2:28:11that's theoretically sound,
- 2:28:13invalid.
- 2:28:13SPEAKER_02: I like all this stuff that we're reading.
- 2:28:17SPEAKER_02: I love this piece that Nick wrote.
- 2:28:19SPEAKER_02: I do.
- 2:28:19SPEAKER_02: I love Bostrom.
- 2:28:21SPEAKER_02: I don't have strong positions on this stuff right now.
- 2:28:26SPEAKER_02: Yeah.
- 2:28:29SPEAKER_02: I feel like I'm still getting acquainted with a lot of this.
- 2:28:33And I ultimately view all of these guys as providing helpful
- 2:28:41speculations on what
- 2:28:44SPEAKER_02: could be coming in the near future.
- 2:28:48And I think each one of these perspectives gives us something
- 2:28:53to think about and
- 2:28:55SPEAKER_02: gives us a different view on things.
- 2:28:58that i mean do do i know like do i think all of this is
- 2:29:02absolutely yeah artificial
- 2:29:04super intelligence is absolutely going to happen like we talked
- 2:29:07about last time
- 2:29:08zero accelerationism james ellis i mean the global economy
- 2:29:12could collapse climate
- 2:29:13change could kick in and everything's wiped out before anything
- 2:29:17uh really happens
- 2:29:19SPEAKER_02: with any of this so i you know i don't
- 2:29:23SPEAKER_02: I don't know what the future holds,
- 2:29:24SPEAKER_02: but I think these types of speculations on whatever AI could become,
- 2:29:32SPEAKER_02: that's how you traffic with the future.
- 2:29:34SPEAKER_02: Not by saying it's necessarily going to be this way,
- 2:29:36but trying to think through these different possibilities that
- 2:29:40we're facing.
- 2:29:41SPEAKER_00: Yeah.
- 2:29:43SPEAKER_00: Yeah, I think something's coming.
- 2:29:50SPEAKER_02: That's what we all feel.
- 2:29:51SPEAKER_00: Right.
- 2:29:51SPEAKER_02: Even my, you know, my coworkers in the warehouse feel that.
- 2:29:54And so I think that's why it's important for us to be having
- 2:29:59these kinds of
- 2:29:59SPEAKER_02: conversations is because everybody normies are feeling the,
- 2:30:05SPEAKER_02: the,
- 2:30:05SPEAKER_02: the outside,
- 2:30:06SPEAKER_02: the,
- 2:30:07SPEAKER_02: Leaking in like leaking in.
- 2:30:09SPEAKER_02: Yeah, they have this premonition in some hazy sense.
- 2:30:14SPEAKER_02: And so, yeah, nobody knows what it's going to be.
- 2:30:17SPEAKER_02: But the point is to try to think through where this could all be going.
- 2:30:21SPEAKER_00: Yeah.
- 2:30:22SPEAKER_00: Well,
- 2:30:23SPEAKER_00: I think,
- 2:30:24SPEAKER_00: I mean,
- 2:30:24SPEAKER_00: at the end of the day,
- 2:30:25SPEAKER_00: the only valid position is that we don't know what's coming,
- 2:30:30SPEAKER_00: but we would like to be prepared to deal with it.
- 2:30:36SPEAKER_00: Yeah, as much as we can be.
- 2:30:38SPEAKER_00: So we are following the Omohundro drives, at least in that much.
- 2:30:46SPEAKER_00: We are trying to be prepared.
- 2:30:51I just think it is a bit absurd to be so certain of doom or of
- 2:30:58utopia,
- 2:30:59of either one.
- 2:31:00SPEAKER_00: I don't think we have the right to be certain.
- 2:31:05Well,
- 2:31:05remember Slavoj Zizek's critique of land is that land is too
- 2:31:10optimistic.
- 2:31:12SPEAKER_02: Because he thinks he knows with some sort of
- 2:31:17SPEAKER_02: I don't think land would say certainty,
- 2:31:18SPEAKER_02: but he,
- 2:31:19SPEAKER_02: he,
- 2:31:19SPEAKER_02: he feels like he knows how this is going to end.
- 2:31:22And Slava was like,
- 2:31:24no,
- 2:31:24we don't have a clue about where this is actually going.
- 2:31:28Um,
- 2:31:30it's important to speculate on it,
- 2:31:31but we must remember our speculations.
- 2:31:34SPEAKER_02: You know, he's a Hegelian.
- 2:31:36SPEAKER_02: So it's going to be the whole, the owl Minerva flies at dusk.
- 2:31:38It's only after whatever happens that we can truly theorize
- 2:31:44about what actually happened.
- 2:31:46SPEAKER_02: Um,
- 2:31:47SPEAKER_02: Until then, we're taking shots in the dark on where this is going.
- 2:31:52SPEAKER_02: But Slavo would say,
- 2:31:54even if you provide an image of where this is going that's
- 2:32:01doomer or pessimistic,
- 2:32:03SPEAKER_02: it's optimistic because it's like,
- 2:32:05SPEAKER_02: okay,
- 2:32:06SPEAKER_02: at least we know it's going to be bad in this way.
- 2:32:08SPEAKER_02: Yep.
- 2:32:10And he's saying we can't even know if it's going to be bad in
- 2:32:12any specific way or not.
- 2:32:15SPEAKER_00: We just got to be in general prepared for what comes next.
- 2:32:19SPEAKER_00: But yeah, it is too optimistic to be certain about anything.
- 2:32:24SPEAKER_00: Don't fall prey to your intuition.
- 2:32:31SPEAKER_00: Other than, you know, we all intuit that something is coming.
- 2:32:37SPEAKER_00: But thinking we know what it's going to be is foolish.
- 2:32:47SPEAKER_00: And on that note, we'll close out this episode.
- 2:32:51SPEAKER_00: We'll be back.
- 2:32:52SPEAKER_00: We'll read these pieces.
- 2:32:55SPEAKER_00: We'll read chapter seven of super intelligence.
- 2:32:59Nick Bostrom will read the diagonal method and discuss that and
- 2:33:06get into more stuff.
- 2:33:08SPEAKER_02: I have a couple other cool things.
- 2:33:10SPEAKER_02: We won't tell them about it, but we got a couple other cool things.
- 2:33:12SPEAKER_02: Oh, yeah.
- 2:33:14SPEAKER_02: In the next couple weeks.
- 2:33:15SPEAKER_00: Lots of cool shit.
- 2:33:17SPEAKER_00: Stay tuned.
- 2:33:18SPEAKER_00: Stay tuned.
- 2:33:18SPEAKER_00: We'll see you guys all soon.
- 2:33:19SPEAKER_00: And don't eat yellow snow or whatever.
- 2:33:24SPEAKER_00: Fucking peace out, guys.
- 2:33:27SPEAKER_00: Word.