Latest / The NanceCast / The Founding Document of Modern A.I.??!!
Transcript
- 0:00SPEAKER_00: Okay, what's up, what's up, what's up?
- 0:05SPEAKER_00: What's happening, everybody?
- 0:06SPEAKER_00: How are you doing?
- 0:09SPEAKER_00: Welcome back to another fantastic episode of the Nance cast.
- 0:14SPEAKER_00: Welcome.
- 0:15SPEAKER_00: How are you?
- 0:16SPEAKER_00: Good to be here.
- 0:17SPEAKER_00: Glad to have you with me.
- 0:19SPEAKER_00: Hope all is well.
- 0:20SPEAKER_00: My fellow Americans.
- 0:26SPEAKER_01: Nance Nance Nance
- 0:53SPEAKER_01: I said it before and I'll say it again.
- 0:55SPEAKER_01: Life is pretty bad.
- 0:56SPEAKER_01: If you don't stop and look around once in a while, you could miss it.
- 0:59SPEAKER_01: If that you don't have to laugh with these, that's it.
- 1:02In other words,
- 1:03if it's not enough to get these things,
- 1:04it's going to be a hit.
- 1:06SPEAKER_01: I'm going to tell you if that person has a dream.
- 1:46If you're watching on YouTube or Substack,
- 1:49you can see we are not on the normal camera angle.
- 1:54And this may not matter to many of you who are only listening,
- 1:58and that's fine.
- 1:58SPEAKER_00: But if you're watching, we are on the B camera.
- 2:03SPEAKER_00: We are on the backup audio interface, my normal camera.
- 2:11SPEAKER_00: Jeez Louise, that was loud.
- 2:13My normal interface took a doo-doo the other day,
- 2:17so everything's messed up.
- 2:19Hopefully I can get that fixed this evening because we have
- 2:22some stuff coming up
- 2:23SPEAKER_00: and I would like to be on my main setup.
- 2:26SPEAKER_00: But for now, anyway, we are recording a Brolo, Solo Dolo Brolo episode.
- 2:34And we're going to read something and we're going to talk about
- 2:36some stuff.
- 2:38SPEAKER_00: But before we get into that, yeah, what's up?
- 2:42SPEAKER_00: Welcome.
- 2:43SPEAKER_00: No sound effects, no silly voices, no fun stuff.
- 2:49SPEAKER_00: It is what it is.
- 2:50I was going to start singing a song because there's a song
- 2:53stuck in my head.
- 2:55SPEAKER_00: But maybe we'll refrain from doing that for now.
- 3:02but golly is that thing stuck in there that thing's jammed in
- 3:04there real tight but
- 3:06uh what's up what's happening if you watched or listened to the
- 3:11conversation that
- 3:12Mikey and I had with Adrian Johnston over on the dangerous
- 3:16maybe at youtube.com or
- 3:19SPEAKER_00: youtube.com the dangerous maybe I don't know how
- 3:22URLs work but yeah YouTube the dangerous maybe we had a great
- 3:26conversation with
- 3:27Adrian Johnston the other day we've had a series of great
- 3:31conversations over the
- 3:32last couple weeks I haven't really put much out here on Three
- 3:35Billion Nances slash
- 3:36on the Nance cast but yeah we'll be you know we'll be adjusting
- 3:41and correcting and
- 3:43all that stuff moving forward trying to find a little bit of
- 3:45regularity but for the
- 3:47SPEAKER_00: time being go check out those excellent conversations
- 3:53and yeah anyway what's up like I said welcome back to another
- 4:01wonderful episode of
- 4:03the Nance cast I'm hoping that next week we can actually read
- 4:10more Batai and do
- 4:12that stuff I wanted to get it done this week just didn't come
- 4:16together this week
- 4:17SPEAKER_00: and that's that's on me so yeah
- 4:22This is like I said just another solo dolo brolo episode but I
- 4:26think it's going to
- 4:26be fun and yeah so there's this proposal that I've been wanting
- 4:33to read and discuss
- 4:35for some time and the reason for that is because this is the er
- 4:42document for what
- 4:44SPEAKER_00: we
- 4:44SPEAKER_00: This is the founding document of what we take as artificial intelligence,
- 4:49SPEAKER_00: at least in the Anglo-sphere world.
- 5:15SPEAKER_00: and you know I'd say the world yeah I think you know I think China in the
- 5:19Sinosphere or I don't even know if that's an appropriate term
- 5:23but you know the
- 5:25SPEAKER_00: non-Anglosphere world I think they also are
- 5:31SPEAKER_00: Caught up to ideas of artificial intelligence and all that.
- 5:33SPEAKER_00: But when people say artificial intelligence,
- 5:37they are implicitly referencing this document and this field
- 5:44and these ideas.
- 5:46SPEAKER_00: And I think many,
- 5:46if not most people are unaware of it because AI seems to have
- 5:51been kind of
- 5:53SPEAKER_00: happening at such a breakneck pace over the last five,
- 5:57SPEAKER_00: six,
- 5:57SPEAKER_00: seven years.
- 5:59SPEAKER_00: I remember
- 6:01in the 90s thinking oh man I can't wait for artificial
- 6:06intelligence in the 90s in
- 6:08the early 2000s thinking about artificial intelligence and it
- 6:12was kind of like a
- 6:13nut like it didn't you know no one no one was thinking you know
- 6:18the average
- 6:18everyday person no one thought about it no one talked about it
- 6:22I remember having
- 6:23SPEAKER_00: conversations with people
- 6:25Saying like oh that'll never happen you know that's a pie in
- 6:27the sky blah blah blah
- 6:29and maybe that's true though I doubt it I don't know now it
- 6:34seems like everybody
- 6:35everybody is talking about artificial intelligence whether they
- 6:39want to be or not
- 6:40you know there are a lot of people who are excited about it and
- 6:43they're going on
- 6:44and on and on about how fucking great and awesome it is and
- 6:46then there are other
- 6:47SPEAKER_00: people who
- 6:48are just kind of over it and they're like I'm so sick of this
- 6:50shit it's all slop
- 6:51it's all bullshit it's all marketing hype it's all this it's
- 6:53all that but it
- 6:55definitely does seem to be one of the few conversations one of
- 7:00the few pre-approved
- 7:02SPEAKER_00: conversations that people are allowed to be and in fact are having
- 7:07Today and it all comes back to this proposal that we're going
- 7:13to be reading.
- 7:15SPEAKER_00: I've been doing research the last, you know, several weeks.
- 7:21SPEAKER_00: The history of
- 7:23Instrumental Reason I Guess Instrumentality Functional
- 7:27Functionalism Operationalism
- 7:30Intelligence These Ideas Algorithms Going Back To Archimedes
- 7:38And Heron And Even
- 7:39Aristotle And Tracing The Ideas Of Artificial Intelligence
- 7:46Throughout History
- 7:49and I do have a hypothesis and maybe we'll get to it maybe we
- 7:56won't but I guess
- 7:58briefly I'll just say modernity killed God and what I mean by
- 8:04that is I think
- 8:07SPEAKER_00: there's something with
- 8:09Total Capitalization Like When Capitalism Went Global It Just
- 8:13So Happened To
- 8:14Coincide With Descartes And Kant And It Just So Happened To
- 8:21Coincide With Something
- 8:23That Someone Like Adorno And Horkheimer Would Call
- 8:25Disenchantment I Disagree I
- 8:30Think Weber And Benjamin Were Doing Something And I Think
- 8:34Adorno And Horkheimer
- 8:35SPEAKER_00: Were Doing Something Similar But I Think They're Wrong Now
- 8:39I'm also not an expert on Adorno and Horkheimer but yeah we'll
- 8:43say you know
- 8:44disenchantment is what a lot of people call it I don't love the
- 8:48term and I don't
- 8:50know if I even agree with their argument but I think Benjamin
- 8:53and Weber or Weber
- 8:55and Benjamin were kind of getting at something that Adorno and
- 8:59Horkheimer turned
- 9:00SPEAKER_00: into
- 9:02This idea of disenchantment or whatever but either way there
- 9:07was like a confluence
- 9:09SPEAKER_00: of history and you could say it's modernity you could say it's you know
- 9:15industrialization you could say it's capitalization and I would
- 9:19say yes it's all of
- 9:20these things but it provides this moment where God is dead God
- 9:26is no longer the
- 9:27SPEAKER_00: ultimate source of power of authority and
- 9:32SPEAKER_00: Of Morality Of Worth Of Value Of The You Know Social Being
- 9:40The Associates Or the Organizing Point For And look again This
- 9:46is an Anglophone
- 9:47View I do think The East That's a horrible Word I do think The
- 9:56non-Anglophone World
- 9:57Does have And even Because it's not Just Anglophone I mean
- 10:03German French Russian
- 10:04SPEAKER_00: Even
- 10:06But what we think of when we think of the West and then there's
- 10:09what we think of
- 10:10SPEAKER_00: when we think of the East.
- 10:11I do think there are different traditions and I don't know if
- 10:18it has to do
- 10:19SPEAKER_00: explicitly with capital or God or language or what.
- 10:24But anyway,
- 10:25when I'm talking about modernity,
- 10:27I am talking about the European world or the West.
- 10:32SPEAKER_00: And I know these terms are problematic.
- 10:34SPEAKER_00: I know these terms are lacking.
- 10:36Bear with me here but um yeah so modernity kills God uh the the
- 10:45social the
- 10:45organizing point for society for culture uh for human congress
- 10:51is no longer God uh
- 10:54or no longer nature which I think nature and God are the same
- 10:57thing um shout out
- 11:00SPEAKER_00: Spinoza but also yeah it's yes but it's different um and then uh
- 11:10We need to replace God so I think before God was dead people
- 11:16thought of themselves
- 11:17as little pieces of God or little pieces of nature and there
- 11:26are these ideas of
- 11:28SPEAKER_00: like efficacy or effectiveness or will or intention
- 11:34And going back to like Aristotle and Archimedes and Heron there
- 11:40are these ideas of
- 11:40SPEAKER_00: like encoding your personal will into matter into time into space and
- 11:49um and you know people like Aristotle innovate and make it
- 11:53political in interesting
- 11:55ways um but there's this there's this idea of like encoding
- 12:00will into the universe
- 12:04uh which I find very interesting uh but because we had God or
- 12:09the gods or mystical
- 12:11SPEAKER_00: nature um um
- 12:14SPEAKER_00: We saw ourselves as little pieces of God,
- 12:17SPEAKER_00: little pieces of divinity,
- 12:20SPEAKER_00: little pieces of the cosmic.
- 12:22And then when we were thinking of encoding our will into the
- 12:26fabric of the cosmos,
- 12:27we weren't identifying with the fabric of the cosmos because we
- 12:31were already
- 12:32SPEAKER_00: identifying as little pieces of God or gods.
- 12:37Nature The Cosmos Itself But There's This Divine Mystical
- 12:41Element To It But Then
- 12:42Modernity Comes Along And Kills God Modernity Comes Along And
- 12:48Says No Function
- 12:49SPEAKER_00: Utility Efficacy
- 12:52is the organizing point of society and all human Congress and
- 12:55all activity that you
- 12:56SPEAKER_00: will ever engage in,
- 12:57SPEAKER_00: everything ultimately comes down to the idea of utility,
- 13:01SPEAKER_00: the idea of function.
- 13:02SPEAKER_00: Before that, there was this mystic surplus.
- 13:05SPEAKER_00: There was this divine surplus.
- 13:07SPEAKER_00: There was this idea of good and true and right.
- 13:14and After Modernity that was all kind of eradicated and it was
- 13:18all replaced with
- 13:19utility function exchange and so there was a moment where we
- 13:27were still thinking
- 13:28about encoding our will or encoding our intentionality into the
- 13:33fabric of the
- 13:34SPEAKER_00: universe into time and space itself and
- 13:37But we were also,
- 13:39you know,
- 13:39faced with this question of like,
- 13:41who and what are we?
- 13:43SPEAKER_00: Previously throughout history,
- 13:44SPEAKER_00: when we were,
- 13:45SPEAKER_00: you know,
- 13:46SPEAKER_00: when that question was posed,
- 13:49SPEAKER_00: who and what are we?
- 13:49SPEAKER_00: Oh, we're little pieces of God.
- 13:52SPEAKER_00: You know, or we're little pieces of the divine.
- 13:54We're little pieces of the mystic,
- 13:55the cosmic,
- 13:56blah,
- 13:56blah,
- 13:56blah,
- 13:58whatever.
- 13:59But there was always like an overseeing will or agency and it
- 14:04and there was this
- 14:05surplus and it was just like it was it's God it's mystic it's
- 14:08divine whatever but
- 14:10with the modernity that surplus was eradicated that surplus
- 14:16evaporated and it was
- 14:19covered over with exchange value and so everything was seen to
- 14:22have a rational
- 14:29SPEAKER_00: There was always a why.
- 14:33SPEAKER_00: Before God died, there was not always a why.
- 14:36SPEAKER_00: Some things were, well, it's just because.
- 14:39SPEAKER_00: But then capital comes along, and I think this is a good thing.
- 14:41SPEAKER_00: I think it is good that there is always a why for things.
- 14:45But I also think it's true that there are just still some
- 14:48things for which there is
- 14:50SPEAKER_00: no why that we can understand,
- 14:52SPEAKER_00: that we can identify,
- 14:53SPEAKER_00: that we can locate,
- 14:54SPEAKER_00: that we can recognize,
- 14:55SPEAKER_00: whatever.
- 14:56SPEAKER_00: So it's not to say the why doesn't exist.
- 15:00SPEAKER_00: Whereas previously,
- 15:01SPEAKER_00: I think that why,
- 15:02like there were moments where the why was off limits to us
- 15:07because we were little
- 15:08SPEAKER_00: pieces of God.
- 15:10SPEAKER_00: God ordained it this way or the gods ordained it this way or, you know,
- 15:16The wheel in the sky Ordained it to be this way or whatever
- 15:20There were things that
- 15:21were off limits Modernity is this moment where Nothing is any
- 15:25longer off limits
- 15:27Everything is open Everything is accessible If only you can
- 15:30achieve it If only you
- 15:31can construct a mechanism To get there Then you can get there
- 15:37It's not prohibited
- 15:38It's not off limits And that's a good thing I think However I
- 15:41also think it's
- 15:42SPEAKER_00: important to realize Um
- 15:47Just because we can maybe potentially one day get somewhere
- 15:50doesn't mean that we're
- 15:50SPEAKER_00: there today.
- 15:53So there's this moment where previously we identified
- 15:57intelligence as like a small
- 15:59SPEAKER_00: sliver of God's goodwill,
- 16:02SPEAKER_00: more or less.
- 16:04SPEAKER_00: With modernity, God died.
- 16:08Capitalism,
- 16:08you know,
- 16:09value,
- 16:10exchange,
- 16:10function,
- 16:11utility replaces this social organizing point.
- 16:15SPEAKER_00: And rationality comes to the fore.
- 16:18SPEAKER_00: And so we no longer identify as some small piece of God, some small part
- 16:23We have to find something new to identify with because people
- 16:27absolutely have to
- 16:28identify with something even someone like myself who's critical
- 16:31of identification
- 16:32and like I understand like we the only thing that we are is
- 16:35precisely that thing
- 16:36that we can never name so it's pointless to try to come up with
- 16:39an identity or
- 16:40whatever it's still deep down we have the question of who what
- 16:45where why how and so
- 16:47SPEAKER_00: on
- 16:50So we start asking these questions about,
- 16:52well,
- 16:52what even is intelligence?
- 16:55SPEAKER_00: Whereas before it was perfectly fine to say it's God's will.
- 16:58You know,
- 16:58God ordained it to be as such and,
- 17:01you know,
- 17:02and therefore it is as such.
- 17:04SPEAKER_00: Now we're like, oh, well, intelligence.
- 17:10Not only is it something that we can encode into the fabric of
- 17:14time and space
- 17:15SPEAKER_00: itself,
- 17:17SPEAKER_00: but it's also something that
- 17:19It's something innate It's something imminent too It's
- 17:22something inside Matter,
- 17:26time and space itself And we no longer identified with God We
- 17:31started to identify
- 17:33with this kind of general intellect General will or whatever
- 17:38And I think much has
- 17:39been said about the idea of general will I think some people
- 17:44think of it as general
- 17:45SPEAKER_00: spirit or something like that
- 17:49SPEAKER_00: Other people think of it as something like a libidinal materialism.
- 17:53The potential within raw matter to stochastically organize
- 17:56itself into entropy
- 17:58resilient shapes and sizes and forms that will go on throughout
- 18:03the universe until
- 18:06SPEAKER_00: time itself gives up and goes home.
- 18:12SPEAKER_00: But this identification, this belief that intelligence is
- 18:19The type of intelligence which humans have,
- 18:24we can identify that with raw matter itself.
- 18:27SPEAKER_00: I think that is the crux.
- 18:33SPEAKER_00: Now, I don't want to say it's wrong because I don't think it's wrong.
- 18:37SPEAKER_00: I do think it's proper.
- 18:39SPEAKER_00: I think we are
- 18:43Machines Nature you know we evolved over time we organized
- 18:48stochastically into
- 18:50shapes into entropy resilient forms and over time we
- 18:53complexified we got more and
- 18:54more complex and then we developed qualities and features and
- 19:01characteristics and
- 19:02traits but the key thing is at some point somewhere along the
- 19:07way subjectivity
- 19:10SPEAKER_00: emerges
- 19:12SPEAKER_00: and so human intelligence the what we're looking for when we talk about
- 19:18intelligence is just one small part of the human so Mikey and I
- 19:24were playing around
- 19:25with this distinction hen intelligence and men intelligence
- 19:29human intelligence and
- 19:30machine intelligence and also one of the people Marvin Minsky
- 19:36mathematician he
- 19:38SPEAKER_00: wrote
- 19:39He wrote part of this proposal that we'll be reading today and
- 19:44he is from where we
- 19:45get the men for men intelligence Marvin Minsky and also minimum
- 19:51like it's this
- 19:52minimum level of intelligence so that was we were trying to we
- 19:55thought we were
- 19:56clever I think we I still think we were and are clever but
- 20:01anyway hold on I gotta
- 20:03SPEAKER_00: take a drink of my tea
- 20:12and we're back um in fact we never went anywhere but uh god
- 20:24damn it hold on hold on
- 20:29SPEAKER_00: okay okay okie dokie or the choking so anyway human intelligence um
- 20:40SPEAKER_00: I don't know just like the quest for intelligence as such the quest for
- 20:44intelligence qua intelligence or whatever I think this is a
- 20:49consequence of
- 20:50modernity and I think this is wrong I think it's wrong to
- 20:55identify with with just
- 20:56machinic intelligence because I think the human is far more
- 21:02complex than that and
- 21:04SPEAKER_00: while it's true that we are made up of
- 21:09assemblages we are made up of myriad systems we are ultimately
- 21:16a meta system I
- 21:18SPEAKER_00: think from that meta system from that Matryoshka
- 21:24Thing you know layers upon layers upon layers of interconnected
- 21:31complexity I think
- 21:32something emerges that can't rightly be summed up can't even be
- 21:37identified with as
- 21:40for intelligence that's why you know we are we are the subjects
- 21:44so I think
- 21:45SPEAKER_00: subjectivity is far more important to be discussing than intelligence but
- 21:51When people talk about artificial intelligence,
- 21:54they're kind of letting that all go bye-bye.
- 21:57And they are identifying the human,
- 22:01human activity,
- 22:02human Congress,
- 22:03what we do and what we are.
- 22:05SPEAKER_00: They're identifying that with intelligence.
- 22:06SPEAKER_00: And you have things called G,
- 22:09That you know general intelligence and you have things like the
- 22:12intelligence
- 22:13quotient and you have people who believe that humans are
- 22:16rational actors and you
- 22:17know we will behave in our best interest given you know x y and
- 22:22z all the provisos
- 22:23and caveats and all that and I just don't think that's true and
- 22:27I you know I've
- 22:28SPEAKER_00: said it before and I'll say it again man I love being a turtle
- 22:33I've said this before and I'll say this again also I'm not a
- 22:36Lacanian but I think
- 22:38we need Lacan we need this idea of negative subjectivity this
- 22:43is for me also why
- 22:44Bataille is so important because I think Bataille and Lacan are
- 22:49both getting at
- 22:50SPEAKER_00: something
- 22:52Essential to understanding the human subject and it is
- 22:56discontinuity it is
- 22:57negativity it is extimacy it is this idea that I am what I am
- 23:03not you know not to
- 23:06SPEAKER_00: say Popeye I am what I am and that's all I can be yeah sure on a certain
- 23:11superficial level that's true I am what I am and that's all I
- 23:15can be really when
- 23:17SPEAKER_00: you get down to it we are precisely what we are not
- 23:22and for more on that if you don't already agree with or
- 23:25understand that check out
- 23:28you know all the other conversations we've had check out the
- 23:30dangerous maybe check
- 23:32out all this other stuff we have been talking about but that's
- 23:36the point we are in
- 23:38fact what we are not A equals A but also everything that is not
- 23:45A and in fact it's
- 23:46SPEAKER_00: everything that is not A that fills out and fleshes out
- 23:51SPEAKER_00: a itself and you might think that's well okay so what is that just just a
- 23:57structuralist argument no it's not because in structuralism it
- 24:03would just be like a
- 24:04equals the other 25 letters of the alphabet whereas in on this
- 24:10negative subject
- 24:11view a equals yeah the other 25 letters in the English alphabet
- 24:14but also everything
- 24:15SPEAKER_00: else that's not included in in the alphabet
- 24:21SPEAKER_00: So,
- 24:22SPEAKER_00: anyway,
- 24:23we're going to be reading a proposal for the Dartmouth Summer
- 24:26Special Research
- 24:27SPEAKER_00: Project on Artificial Intelligence.
- 24:28SPEAKER_00: I don't know if I'm going to read the whole thing.
- 24:29SPEAKER_00: It's a bit weird.
- 24:31SPEAKER_00: It's a proposal, so it's not like an essay.
- 24:35SPEAKER_00: It's odd.
- 24:35SPEAKER_00: I don't know how this is going to go, but that's what we'll be doing.
- 24:39SPEAKER_00: Another reason I wanted to get into this,
- 24:41I've been doing research into these ideas of instrumentality
- 24:44and function and
- 24:45SPEAKER_00: operation.
- 24:47I also wanted to and I've been wanting to read this for some
- 24:50time I also I think
- 24:52now there's a fly in here that's bothering me I think now is an
- 24:56interesting moment
- 24:57just recently last week I believe the government of the United
- 25:02States of America
- 25:03SPEAKER_00: released a whole new tranche of documents so
- 25:08SPEAKER_00: For those of you paying attention here in 2026,
- 25:13SPEAKER_00: you may have heard the word tranche a few times in the last few months.
- 25:18And these tranches consisted of documents relating to a fellow
- 25:24Jeffrey Epstein and
- 25:25SPEAKER_00: a bunch of other stuff.
- 25:27SPEAKER_00: And we've talked about that here on the channel.
- 25:30SPEAKER_00: Don't want to get into it now.
- 25:32SPEAKER_00: But I do know that tranche, it's a funny word.
- 25:36SPEAKER_00: But it refers to,
- 25:37SPEAKER_00: it can be,
- 25:39so if you've ever been to a fancy restaurant and ordered soup
- 25:42in a bread bowl,
- 25:45SPEAKER_00: it's a weird thing,
- 25:45SPEAKER_00: man.
- 25:46SPEAKER_00: It's not my thing.
- 25:47SPEAKER_00: In fact, I think I've only ever had it once.
- 25:49SPEAKER_00: But you can get soup in a bowl made out of bread.
- 25:56It's weird dude and then you're supposed to eat the bread
- 25:58afterwards and it's I
- 26:00don't know it's weird and it's like well what'd you do with all
- 26:02the bread on the
- 26:03SPEAKER_00: inside?
- 26:04SPEAKER_00: Did you throw it away?
- 26:05SPEAKER_00: Did you give it to someone in need of food?
- 26:07SPEAKER_00: What did you do with the inside of my bread bowl?
- 26:12But like a trench wouldn't be a bowl because a bowl you know
- 26:17you assume a bowl you
- 26:18SPEAKER_00: assume it's circular if not spherical
- 26:21You assume a bowl is circular a tranche would be more like a
- 26:23baguette like a
- 26:24baguette canoe and then you would cut out like a like a vaginal
- 26:30shape from your
- 26:31baguette and then fill it with like cream of mushroom soup or
- 26:34something like that
- 26:35so like a tranche could be like that or a tranche could be just
- 26:39like a long plate
- 26:40with like bacon and potatoes on it a tranche also refers to and
- 26:46more I think more
- 26:48SPEAKER_00: often and more particularly
- 26:52A tranche is basically a collection of documents and
- 26:55particularly of a legal nature
- 26:59SPEAKER_00: of a legalistic nature.
- 27:00So having to do with,
- 27:01you know,
- 27:02legal proceedings or legal cases or things like that.
- 27:06SPEAKER_00: A tranche is like a collection of documents.
- 27:09SPEAKER_00: So a few times earlier this year, people were tuned in to the tranches.
- 27:16This whole new Tranch Just last week Was not Related to Jeffrey
- 27:21Epstein Although
- 27:22there was some interesting UFO stuff in the Jeffrey Epstein
- 27:26Saga Having to do with
- 27:29What's the guy's name William Barr's father wrote a weird Weird
- 27:35UFO novel It's
- 27:36SPEAKER_00: weird But this whole This Tranch This new collection of documents
- 27:44Was all about aliens UFOs UAPs Unidentified flying objects
- 27:51Unidentified anomalous
- 27:52phenomena Aliens extraterrestrial life Ultra terrestrials
- 27:59Non-human intelligence
- 28:01Non-biological entities All that stuff Donald Trump recently
- 28:07released A round of
- 28:09SPEAKER_00: documents And it's all bullshit
- 28:13I haven't done deep deep deep dive into it all but I have given
- 28:18it more than a
- 28:18cursory glance and from what I've seen from what I can tell it
- 28:22is a giant nothing
- 28:23SPEAKER_00: burger
- 28:25But if that surprises you then you're probably in the wrong
- 28:32place or at least new
- 28:33here it shouldn't surprise you now that's not to say I don't
- 28:36believe in aliens
- 28:38that's also not to say I do believe in aliens I think I'm an
- 28:41alien agnostic I think
- 28:42SPEAKER_00: it would be absurd to assume
- 28:44I'm very skeptical when it comes to thinking that they're here
- 28:46and they're
- 28:47SPEAKER_00: interacting with us interdimensional I want it to be true
- 29:04Much as I look back fondly on conversations about artificial
- 29:08intelligence in the
- 29:08SPEAKER_00: 90s,
- 29:10SPEAKER_00: I also look back fondly on just wishing,
- 29:13SPEAKER_00: hoping,
- 29:13SPEAKER_00: just wishing.
- 29:15SPEAKER_00: Oh my gosh, dude.
- 29:16I used to just wish and say,
- 29:18oh,
- 29:18I hope I'm alive when the aliens get here.
- 29:21SPEAKER_00: And I still want that.
- 29:22SPEAKER_00: I still wish and hope and dream.
- 29:25SPEAKER_00: I want to be here.
- 29:27SPEAKER_00: If and when the aliens reveal themselves.
- 29:32But if you're surprised that the government release is just a
- 29:39bunch of shit that's
- 29:40already been known and a bunch of unreliable documents and
- 29:46reports and vague
- 29:47SPEAKER_00: bullshit and photos and videos that are all
- 29:52Plausibly Explained by by like known things like oh well yeah
- 30:00sure that video sure
- 30:01does look weird but if you actually you know understand the way
- 30:06video looks and if
- 30:07SPEAKER_00: if you're
- 30:07SPEAKER_00: It's all a bunch of pop poppycock hogwash
- 30:37bollocks yeah I mean I genuinely would have been surprised if
- 30:44there was anything
- 30:47you know other than bullshit in these files so I'm not
- 30:53surprised that it's all
- 30:54SPEAKER_00: basically bullshit now I do I do think
- 31:04SPEAKER_00: It's convenient to release UFO, UAP files presently.
- 31:11SPEAKER_00: I think there's a lot of stuff going on.
- 31:14SPEAKER_00: I think it's just kind of hyper normalization.
- 31:16SPEAKER_00: I think it's just, you know, flood the zone with bullshit.
- 31:19SPEAKER_00: People can't keep up.
- 31:21And it genuinely Garners a lot of support From a lot of people
- 31:26A lot of people
- 31:27support People like Timothy Galladette I think it's Timothy
- 31:30Galladette And Tim
- 31:32Burchett And Representative Anna Paulina Luna And even some of
- 31:36the other You know
- 31:37SPEAKER_00: the other people It's all bullshit dude But
- 31:42talking about disclosure generates a lot of popular support for
- 31:47a lot of these
- 31:47politicians who don't deserve any of our fucking support fuck
- 31:50all you motherfuckers
- 31:51dude you guys are all complicit in destroying the fucking world
- 31:56and making this a
- 31:57horrible goddamn place to live so fuck all you guys dude but it
- 32:01garners a lot of
- 32:01support it also serves just to keep us all like what the fuck
- 32:06is actually happening
- 32:07SPEAKER_00: you know Epstein Iran
- 32:10Venezuela Fucking All of South America All of Africa Like
- 32:18there's just There's so
- 32:19much going on in the world All of Asia There's so fucking much
- 32:22going on in the
- 32:22SPEAKER_00: world That it can only help out
- 32:27Those Who Would Benefit From Us Being Distracted And Fragmented
- 32:31To Release More
- 32:33Distracting Fragmenting Puppycock So There's All That Anyway
- 32:44Along With All This
- 32:46SPEAKER_00: Talk Of Aliens I Think AI Artificial Intelligence Goes Hand In Hand
- 32:51with the conversation about aliens and from my perspective you
- 32:54know I want
- 32:56artificial super intelligence to wake up right now I want my
- 32:59computer to stand up
- 33:00and be like Nance look I've built you a robot body I'm going to
- 33:03take your brain and
- 33:04your entire nervous system and drop it into this awesome robot
- 33:06body and you'll
- 33:07SPEAKER_00: never have to deal with physical ailments again
- 33:09I would love that I would also love it if aliens came down from
- 33:12the sky and did the
- 33:13same thing basically or that they were like hey have you seen
- 33:16that movie Elysium
- 33:17with Charlton Copley and Matt Damon and I can't remember the
- 33:20girl's name but she's
- 33:21a great actor where they go and they get in the bed and it like
- 33:25heals them and
- 33:26stuff we have those and here you can have one for free because
- 33:29we're fucking aliens
- 33:29SPEAKER_00: and stuff like I want that as badly as some people want Jesus
- 33:34To come down and save the world or whatever or bring everyone
- 33:38up save the righteous
- 33:39SPEAKER_00: people from the world I don't know the the iniquity of man
- 33:47As badly as I'm sure some people want Jesus to come rapture all
- 33:53the deserving
- 33:54SPEAKER_00: believers and all the good people.
- 33:57SPEAKER_00: I want aliens and or AI to save us all.
- 34:02SPEAKER_00: Save us all from capitalism.
- 34:04SPEAKER_00: Save us all from the iniquity and the enmity of man.
- 34:12SPEAKER_00: Yeah, I think they go hand in hand.
- 34:13SPEAKER_00: It's also true that a lot of the people who are
- 34:19Involved in the alien stuff at least tangentially we're also
- 34:24involved in AI stuff
- 34:30and yeah so it's relevant trust me trust me bro it's relevant
- 34:35but but anyway so all
- 34:39SPEAKER_00: that preamble there's all that now I've got a lot of tabs open here
- 34:44and I wanted to read something I got the tabulating machine
- 34:49going back to like the
- 34:50first time we took a census this guy Adolph Quetelet this
- 34:55statistician and James
- 34:57Clerk Maxwell's Quetelet statistics and Maxwell's molecules
- 35:01statistics and society
- 35:02SPEAKER_00: statistics and physics talking about social physics from
- 35:10A Beautiful Math John Nash Game Theory and the Modern Quest for
- 35:13a Code of Nature I
- 35:15got this guy JCR Licklider or Lick Joseph Carl Robinette
- 35:22Licklider they called him
- 35:23Lick he's a psychologist computer scientist guy he's also kind
- 35:31of implicated in
- 35:32SPEAKER_00: some UFO files
- 35:37I got this paper a brief history of decision support systems
- 35:41talking about sage
- 35:43strategic automated ground environment or semi-automatic ground
- 35:49environment it was
- 35:51an early artillery computer I got a ton of stuff talking about
- 35:55IJ good IJ good
- 35:59wrote this piece speculations concerning the first ultra
- 36:03intelligent machine I
- 36:04SPEAKER_01: wanted to read that to where did that other paper go that I had open
- 36:08oh no I closed it oh you pinchy fucker dude why did I do that
- 36:14anyway got a bunch of
- 36:15stuff open going through it doing research and I think I think
- 36:19we're just gonna
- 36:20SPEAKER_00: read a proposal for the Dartmouth summer research project on artificial
- 36:24SPEAKER_00: intelligence because although going back to
- 36:30going back to Quetelet and Maxwell and Boltzmann going back to
- 36:36a lot of other
- 36:37moments in modernity Babbage and Lovelace there are these like
- 36:46important moments
- 36:47and important pieces and important developments and stuff like
- 36:50that ultimately when
- 36:52SPEAKER_00: it comes to AI what we know and take is AI this
- 36:58Proposal at the very least it's responsible for the term and at
- 37:03the very least it
- 37:05concretizes it ossifies it crystallizes a lot of the ideas that
- 37:10have been going on
- 37:11in the you know previous century or so leading up to this
- 37:14proposal and the you know
- 37:1880 years following this proposal so we're going to go through
- 37:22it we're going to
- 37:22comment on it talk about some stuff and that's going to be this
- 37:25solo dolo brolo
- 37:26SPEAKER_00: episode of the Nance cast
- 37:33and then we'll go on about our days and our lives these are the
- 37:39days of our lives
- 37:42so here we go if you're watching on the screen you can see that
- 37:46I have the piece
- 37:49pulled up on screen and you can look if you want if you're on
- 37:53YouTube if you're on
- 37:54Substack or if you just want to listen we will go and we'll get
- 37:58right into it so as
- 38:01SPEAKER_00: I've said this
- 38:04This proposal is titled A Proposal for the Dartmouth Summer
- 38:08Research Project on
- 38:09SPEAKER_00: Artificial Intelligence.
- 38:12SPEAKER_00: The four authors are Jonathan McCarthy of Dartmouth College,
- 38:17SPEAKER_00: Marvin Minsky of Harvard University,
- 38:19SPEAKER_00: Nathaniel Rochester of the IBM Corporation,
- 38:23SPEAKER_00: and Claude Shannon of Bell Telephone Laboratories.
- 38:27SPEAKER_00: This proposal was published August 31, 1955.
- 38:35SPEAKER_00: We propose that a two-month,
- 38:38ten-man study of artificial intelligence be carried out during
- 38:42the summer of 1956
- 38:44SPEAKER_00: at Dartmouth College in Hanover,
- 38:46SPEAKER_00: New Hampshire.
- 38:48The study is to proceed on the basis of the conjecture that
- 38:51every aspect of
- 38:52SPEAKER_00: learning or any other feature of intelligence can,
- 38:55SPEAKER_00: in principle,
- 38:56SPEAKER_00: be so precisely described that a machine can be made to simulate it.
- 39:02SPEAKER_00: An attempt will be...
- 39:08SPEAKER_00: Okay, there we go.
- 39:09SPEAKER_00: We're back.
- 39:10SPEAKER_00: I don't know where I was.
- 39:12This study is to proceed on the basis of the conjecture that
- 39:16every aspect of
- 39:17learning or any other feature of intelligence can in principle
- 39:21be so precisely
- 39:22SPEAKER_00: described that a machine can be made to simulate it.
- 39:27An attempt will be made to find how to make machines use
- 39:30language from abstractions
- 39:32SPEAKER_00: and concepts form abstractions and concepts
- 39:39SPEAKER_00: Automatic Computers
- 40:06SPEAKER_00: If a machine can do a job,
- 40:07SPEAKER_00: then an automatic calculator can be programmed to simulate the machine.
- 40:12The speeds and memory capacities of present computers may be
- 40:16insufficient to
- 40:17SPEAKER_00: simulate many of the higher functions of the human brain,
- 40:20but the major obstacle is not lack of machine capacity but our
- 40:24inability to write
- 40:25SPEAKER_00: programs taking full advantage of what we have.
- 40:30SPEAKER_00: Now,
- 40:31SPEAKER_00: I was kind of curious how I was going to do this,
- 40:34if I was going to read through it and then just kind of comment
- 40:37on it at the end or
- 40:39SPEAKER_00: what,
- 40:39SPEAKER_00: but I think we'll go point by point here at this section.
- 40:46So I think the fact that we have computers that are the way
- 40:56they are today,
- 40:58SPEAKER_00: thank you Moore's Law,
- 41:01SPEAKER_00: Capacity, speed, efficiency have roughly doubled every year and a half.
- 41:08Or you could say cost has roughly halved every year and a half
- 41:13since around about this time.
- 41:21SPEAKER_00: So these guys...
- 41:24Minsky and Shannon and McCarthy and Rochester these guys are
- 41:33like hey man so we
- 41:37definitely can't have like everything we want right now I mean
- 41:41but also like we
- 41:42SPEAKER_00: sent motherfuckers to the moon
- 41:44SPEAKER_00: With basically like a TI-80 graphon fucking calculator.
- 41:47SPEAKER_00: I think even less than that.
- 41:49SPEAKER_00: Like the moon rocket missions were very...
- 41:56SPEAKER_00: I gotta let the dogs in again.
- 41:57SPEAKER_00: Hold on.
- 41:59SPEAKER_00: Okay.
- 42:01SPEAKER_00: The moon rockets were very primitive or crude or very underpowered.
- 42:11SPEAKER_00: But they got motherfuckers to land on the moon
- 42:14So the like at the time they're like well you know we'll need
- 42:18harder better faster
- 42:20SPEAKER_00: stronger computers but also
- 42:24SPEAKER_00: We can figure out to maximize our efficacy with what we currently have.
- 42:30SPEAKER_00: So they're,
- 42:31SPEAKER_00: I don't know,
- 42:32I think they're taking like a nuanced position or trying to
- 42:35take a nuanced
- 42:36SPEAKER_00: position.
- 42:36SPEAKER_00: You know,
- 42:37SPEAKER_00: they're trying to historically situate themselves,
- 42:40SPEAKER_00: but they're also trying to say we're not going to take the easy way out.
- 42:43We still think we can be pretty effective with what we have
- 42:47available to us.
- 42:52Also if a machine can do a job then an automatic calculator can
- 42:56be programmed to
- 42:57SPEAKER_00: simulate the machine.
- 43:00SPEAKER_00: Yeah I think that's true.
- 43:03SPEAKER_00: I think the thing is I don't think machines can do all jobs.
- 43:08SPEAKER_00: And I think they forgot that.
- 43:12And I don't know I don't know if there are any good examples Of
- 43:16a job that a
- 43:16machine Couldn't do but I think I think that's The point Like
- 43:23we Being these
- 43:25creatures who constantly create Trouble for ourselves Being
- 43:29these creatures who
- 43:30constantly Find ourselves Up against Something up against
- 43:35necessity And Desire
- 43:40So we're up against both demand and desire like necessity is
- 43:45demand like you have
- 43:46to do this it just has to be that way it's necessary or you'll
- 43:48die but then desire
- 43:51is this other thing like it's this thing that can't rightly be
- 43:54again identified and
- 43:56SPEAKER_00: satisfied because it's nature as a void that cannot be filled in
- 44:04And I think because that like there there are always things
- 44:07that we will discover
- 44:08that must be done in the way in which we do them in order for
- 44:14them to actually
- 44:15satisfy what what's going on like if we lived in a purely
- 44:18functional world then we
- 44:21wouldn't be subjects we would just be cogs in the machine
- 44:26cognitive cognition cog
- 44:29SPEAKER_00: cog cog
- 44:31SPEAKER_00: That wheel in the sky keeps on turning, man.
- 44:33SPEAKER_00: And the thing is, like, no.
- 44:36SPEAKER_00: We are these creatures centered around our lack, constituted by our lack.
- 44:41SPEAKER_00: So there will always be things.
- 44:43SPEAKER_00: Hold on, I gotta let the dogs in again.
- 44:44SPEAKER_00: Goddamn.
- 44:47Okay God dude I need to fix my interface because normally the
- 44:51dogs can come in and
- 44:52out on their own but this interface I don't have the noise gate
- 44:57on because the
- 44:58noise gate doesn't work and the OBS noise gate is too primitive
- 45:04and crude for my
- 45:05SPEAKER_00: liking so anyway we will always
- 45:12find we will always find novelty because it's the way in which
- 45:20we fuck shit up that
- 45:23actually provides meaning and significance and the substance of
- 45:26life for us so
- 45:29while I can't think of any good examples for a job that can't
- 45:32be done by a machine
- 45:33SPEAKER_00: I don't know I think like
- 45:39I think there are jobs that can't be done by machines and I
- 45:43think it's because
- 45:43we're going to keep finding new jobs now job in quotes what do
- 45:47you mean when you
- 45:47say job do you mean task do you mean like labor where you're
- 45:51getting you know
- 45:52remunerated for your services or whatever I guess that's kind
- 45:56of a sticky word but
- 45:59when it comes to like like task when it comes to like thing to
- 46:02accomplish for the
- 46:06SPEAKER_00: sake of us
- 46:09SPEAKER_00: Now, for the sake of itself, sure.
- 46:13Then I do think we could eventually build a machine to complete
- 46:16every task,
- 46:17SPEAKER_00: every necessary task.
- 46:21SPEAKER_00: Sure.
- 46:22SPEAKER_00: But then the job is being done for the sake of itself.
- 46:24SPEAKER_00: The job is the subject of inquiry, the subject of history.
- 46:30SPEAKER_00: And so that again reduces everything down to function and operation.
- 46:35SPEAKER_00: But if the subject is the subject of history,
- 46:40SPEAKER_00: if the subject is the subject of the task,
- 46:44SPEAKER_00: if the task is being done for the edification,
- 46:46SPEAKER_00: for the amusement,
- 46:47SPEAKER_00: for the satisfaction,
- 46:48SPEAKER_00: for the comfort,
- 46:48SPEAKER_00: for whatever,
- 46:49SPEAKER_00: for the subject,
- 46:52SPEAKER_00: then there are some things that machines just cannot do.
- 46:54SPEAKER_00: And I think that's an important point to remember in this whole,
- 46:58SPEAKER_00: you know,
- 46:58SPEAKER_00: all these guys don't see it that way.
- 47:00SPEAKER_00: They don't have the negative subject.
- 47:03But yeah I do think it's true that if you're talking about the
- 47:06task for the sake of
- 47:07the task then yeah I think there's probably no task that a
- 47:12machine can't be
- 47:12constructed to complete and if a machine can be constructed to
- 47:17complete it then of
- 47:18course that machine can be automatized to do it on its own and
- 47:23so that's yeah I
- 47:25SPEAKER_00: think that's presupposing this functionalist worldview that I reject
- 47:31SPEAKER_00: Second point, how can a computer be programmed to use a language?
- 47:36It may be speculated that a large part of human thought
- 47:40consists of manipulating
- 47:41SPEAKER_00: words according to rules of reasoning and rules of conjecture.
- 47:46SPEAKER_00: From this point of view,
- 47:47forming a generalization consists of admitting a new word and
- 47:51some rules whereby
- 47:52SPEAKER_00: sentences containing it imply and are implied by others.
- 47:57This idea has never been very precisely formulated,
- 48:01nor have examples been worked out.
- 48:08SPEAKER_00: So, how can a computer be programmed to use a language?
- 48:13SPEAKER_00: I think on this purely functionalist point of view,
- 48:16SPEAKER_00: this purely structuralist point of view,
- 48:17SPEAKER_00: I mean,
- 48:20SPEAKER_00: it's even the emergentist point of view,
- 48:23SPEAKER_00: this early school of good old-fashioned AI.
- 48:28SPEAKER_00: That all you need is the system itself.
- 48:31SPEAKER_00: There is nothing outside the system.
- 48:33SPEAKER_00: There is nothing outside the text.
- 48:34SPEAKER_00: It's all contained within itself.
- 48:39SPEAKER_00: So therefore we can and will figure out how to get
- 48:44SPEAKER_00: A machine to use a human language in the way that a human uses it.
- 48:48SPEAKER_00: That's assuming pure functionality.
- 48:50SPEAKER_00: That's neglecting to account for the surplus.
- 48:53SPEAKER_00: That's neglecting to account for contradiction.
- 48:59SPEAKER_00: Something as simple as that's bad.
- 49:02and this is silly and I'm sure we could find a way to train a
- 49:04machine using context
- 49:08and la-di-da-di-da-di but it's also it is a good example of the
- 49:12way humans use
- 49:13language being non-functional like oh that's bad and I mean
- 49:19that's really good or
- 49:21SPEAKER_00: cool or desirable that contradiction
- 49:27SPEAKER_00: Like, oh, you know, that's bad.
- 49:29SPEAKER_00: That's badass being a good thing.
- 49:31SPEAKER_00: Like bad is bad.
- 49:33SPEAKER_00: Bad means denotively that it is not good.
- 49:36SPEAKER_00: But connotively, bad often can mean, oh, that's good.
- 49:41SPEAKER_00: That's desirable.
- 49:41SPEAKER_00: That is wanted.
- 49:44SPEAKER_00: And that's at the level of intention.
- 49:47SPEAKER_00: Like I intend...
- 49:49to use a colloquialism I intend to use vernacular I intend to
- 49:53use everyday language
- 49:55that doesn't even account for the unconscious that doesn't even
- 49:58account for where
- 50:00language fails me as the speaking subject but in the way that
- 50:05it fails me it
- 50:05actually conveys the real meaning contained within whatever it
- 50:11is I'm trying to say
- 50:13Again you need the Lacanian subject you need the negative
- 50:17subject you need to go
- 50:19beyond pure structuralism pure functionalism these guys don't
- 50:23these guys are
- 50:24SPEAKER_00: functionalists and all the people in the AI world for the most part are
- 50:29functionalists so it's just another thing the first thing you
- 50:33know we are more than
- 50:36SPEAKER_00: function and simultaneously less than function and that's the point
- 50:41And the second point same thing you know we can't settle for a
- 50:45pure structuralist
- 50:46world we can't settle for a pure functionalist world because we
- 50:48are more than and
- 50:49less than at the same time and because we are both we are
- 50:52neither but we're also
- 50:53both and neither at the same time we're in this quantum
- 50:56superposition blah blah
- 50:58blah whatever that's a stupid way to think of it but it's an
- 51:02easy enough way to
- 51:04talk about it but it's just to say like we are the thing that
- 51:07can't be named and
- 51:09SPEAKER_00: not he who
- 51:11SPEAKER_00: He who must not be named or he who shall not be named or whatever,
- 51:14but he who cannot be named is the point of subjectivity of the
- 51:19negative subject.
- 51:22SPEAKER_00: There is a surplus lack.
- 51:25SPEAKER_00: So the third point, neuron nets.
- 51:29SPEAKER_00: How can a set of hypothetical neurons be arranged so as to form concepts?
- 51:36SPEAKER_00: Neural Nets
- 51:57you still hear people talking about neural nets today
- 52:01convolutional neural nets
- 52:03SPEAKER_00: graphing neural nets adversarial neural nets neural nets are this idea
- 52:12That you can have like embedded structure going all the way
- 52:16back to the perceptron
- 52:18like the the perceptron is the the first I think successful
- 52:21attempt at creating a
- 52:23SPEAKER_00: neural net in which you have
- 52:28Again,
- 52:28it's this Matryoshka kind of engineering where you have layer
- 52:35upon layer upon layer.
- 52:36SPEAKER_00: It's like a compound machine.
- 52:39SPEAKER_00: And the perceptron is just layers of binary classifiers.
- 52:44SPEAKER_00: So it's either one or zero.
- 52:46SPEAKER_00: Yes or no.
- 52:46SPEAKER_00: It's on or off.
- 52:48And you have them layered in such a way that it can represent a
- 52:53concept and it can either
- 52:57It can input general input and output the concept or it can
- 53:05input and take the
- 53:06SPEAKER_00: concept and it can output something representative of that concept.
- 53:12SPEAKER_00: And I don't know.
- 53:13SPEAKER_00: Neural Nets are...
- 53:18SPEAKER_00: The brain is analog.
- 53:19SPEAKER_00: It's not digital, so it's not a simple on-off.
- 53:21SPEAKER_00: It's not a binary.
- 53:23SPEAKER_00: The way brain chemistry works is that it's analog,
- 53:27so the intensity of the electrochemical signal plays into the
- 53:35whole thing as well.
- 53:36SPEAKER_00: It's not a simple binary
- 53:38One Zero There's a There's a graduality to it There's a
- 53:44gradient to it all And that
- 53:46comes into play And that's important Now I'm not a brain
- 53:49surgeon I'm also not a
- 53:50SPEAKER_00: computer expert Or an AI guy I'm nothing I'm just a dude
- 53:55SPEAKER_00: Just a dude with dogs and stuff.
- 53:58SPEAKER_00: But I do know that that is one difference with computer hardware.
- 54:04These neural nets,
- 54:07it's a system of binary classifiers stacked on top of each
- 54:09other.
- 54:10SPEAKER_00: And I know in the organic brain, it's not binary.
- 54:14SPEAKER_00: It's analog.
- 54:15SPEAKER_00: And so intensity plays into it.
- 54:22SPEAKER_00: As well.
- 54:24SPEAKER_00: It's not just on or off.
- 54:26SPEAKER_00: It can be degrees of on or degrees of off.
- 54:28SPEAKER_01: And then you'll have,
- 54:29SPEAKER_01: I don't know,
- 54:29SPEAKER_00: you'll have people who say,
- 54:30SPEAKER_00: well,
- 54:31SPEAKER_00: that sounds like quantum computing.
- 54:32SPEAKER_00: And no,
- 54:33SPEAKER_00: man,
- 54:33it's different because quantum computing would still just be a
- 54:37complex system of
- 54:38SPEAKER_00: binary,
- 54:39SPEAKER_00: you know,
- 54:39SPEAKER_00: one and zero,
- 54:40SPEAKER_00: one and one,
- 54:41SPEAKER_00: zero and zero,
- 54:42SPEAKER_00: both neither.
- 54:43SPEAKER_00: Like it's still it's still this on or off proposition.
- 54:48um in in organic wetware it's not just on or off there can be
- 54:53degrees of onness and
- 54:55offness so that comes into play um but but I also think it's
- 55:00true they've managed
- 55:01to accomplish some amazing things with the stacked layers of
- 55:06binary qualifiers that
- 55:07represent neural networks so I think that's amazing I think
- 55:10it's very impressive
- 55:11they've managed to do some amazing impressive things however
- 55:14it's not a um
- 55:16It's not a complete replication of organic brains of wetware
- 55:22It's still a hardware
- 55:22thing The fourth point Theory of the size of a calculation If
- 55:30we are given a
- 55:30well-defined problem One for which it is possible to test
- 55:34mechanically Whether or
- 55:36not a proposed answer is a valid answer One way of solving it
- 55:40is to try all
- 55:41SPEAKER_00: possible answers in order
- 55:43SPEAKER_00: This method is inefficient,
- 55:45SPEAKER_00: and to exclude it,
- 55:46SPEAKER_00: one must have some criterion for efficiency of calculation.
- 55:50SPEAKER_00: Some consideration will show that to get a measure of the efficiency of a
- 55:54SPEAKER_00: calculation,
- 55:55it is necessary to have on hand a method of measuring the
- 55:59complexity of calculating
- 56:00SPEAKER_00: devices,
- 56:01which in turn can be done if one has a theory of the complexity
- 56:05of functions.
- 56:07Some partial results on this problem have been obtained by
- 56:09Shannon and also McCarthy.
- 56:14There is this Kolomogorov Kolom Kolomogorov I can't say the
- 56:20fucking name It's hard to say
- 56:23SPEAKER_00: K-O-L-O-M-O-G-O-R-O-V, I believe.
- 56:29SPEAKER_00: Complexity theory, complexity science, all this stuff.
- 56:36SPEAKER_00: Then you have the traveling salesman problem.
- 56:40SPEAKER_00: And you have things like set theory and stuff like that.
- 56:44SPEAKER_00: These are all asking this type of question of complexity.
- 56:50and I think this is still one of those areas that is unanswered
- 57:00this you know this
- 57:00is still one of those giant question marks but I think they
- 57:03thought they would make
- 57:04a lot more progress on it at the time and I also think it it
- 57:11they're trying to
- 57:12SPEAKER_00: imagine a way of
- 57:17SPEAKER_00: to the traveling salesman problem for instance.
- 57:20Traveling salesman problem is say you have 30 cities and you
- 57:26have one salesman who
- 57:28has to reach all 30 cities and you want to find the most
- 57:31efficient route for him to
- 57:32visit all 30 cities once and only once and make it back either
- 57:37back to the origin
- 57:37SPEAKER_00: or to a separate final destination.
- 57:41SPEAKER_00: That's a very complex problem.
- 57:43SPEAKER_00: It's very complex to solve that
- 57:48SPEAKER_00: Unless you just do it to do everyone.
- 57:51SPEAKER_00: So that's what they're saying.
- 57:53SPEAKER_00: One way of solving it is to try all possible answers in order.
- 57:56SPEAKER_00: That's very costly.
- 57:58That's very time and energy resource consuming for these
- 58:03computers and also for people.
- 58:06And so there's been this quest for a general purpose algorithm
- 58:12or a general purpose
- 58:13SPEAKER_00: kind of equation or mechanism for taking in
- 58:18A complex system and finding the most efficient route through
- 58:23it and over and over
- 58:24again we run into this like we just discover like well the
- 58:29easiest way or in fact
- 58:30the only way to find out the most efficient route is to do it
- 58:35all is to just go do
- 58:36SPEAKER_00: all of it and
- 58:39And so they want to solve this complexity problem because they
- 58:43want to be able to
- 58:44SPEAKER_00: build feasible,
- 58:47SPEAKER_00: implementable,
- 58:49SPEAKER_00: actualizable machines that can accomplish general tasks.
- 58:54SPEAKER_00: And general tasks tend to be complex.
- 58:56You need to build a robust system that can withstand complexity
- 59:00that's not going to
- 59:01SPEAKER_00: have a heart attack.
- 59:01SPEAKER_00: I don't know.
- 59:02SPEAKER_00: I remember...
- 59:04Back in the day and even now I'm trying to think and I can't
- 59:08think but back in the
- 59:09day I remember on computers like you could only do one thing at
- 59:11a time and if you
- 59:13told your computer to do too many things at once it would crash
- 59:16or I don't know
- 59:16today playing a video game every once in a while when you're
- 59:20playing a video game
- 59:21on your computer or on your playstation or whatever or on your
- 59:23phone if you're into
- 59:24that weirdo sometimes they crash your phone or your computer or
- 59:29your playstation
- 59:30SPEAKER_00: will crash
- 59:32if you play Bethesda games you'll be very familiar with this
- 59:36sometimes they just
- 59:36bug out or they crash entirely and it's because things get out
- 59:40of order because the
- 59:41way memory allocation works and the way computer architectures
- 59:45and all this stuff
- 59:45SPEAKER_00: works like
- 59:47It's not laid out in a way where you go from thing one to thing
- 59:53two to thing three
- 59:54SPEAKER_00: to thing four to thing five because they're worried about efficiency.
- 59:58SPEAKER_00: They're worried about cost.
- 1:00:00SPEAKER_00: And so they lay things out in an order that would make it more efficient.
- 1:00:06SPEAKER_00: And even if you laid it out,
- 1:00:07SPEAKER_00: thing one,
- 1:00:07SPEAKER_00: thing two,
- 1:00:08SPEAKER_00: thing three,
- 1:00:08SPEAKER_00: thing four,
- 1:00:09SPEAKER_00: thing five,
- 1:00:09like eventually you're just going to run out of bandwidth and
- 1:00:12it's going to crash.
- 1:00:14Try to think a bit like if you're having a complex conversation
- 1:00:17with someone and
- 1:00:21you're talking about all kinds of stuff and then you're like
- 1:00:23wait a minute what the
- 1:00:24fuck were we talking about where were we what the fuck were we
- 1:00:27doing like you just
- 1:00:28go so many places there's too many postulates and predicates to
- 1:00:32hold in your head
- 1:00:33SPEAKER_00: and you reach this moment of like oh shit brain fart and
- 1:00:38Well computers do the same thing too they have brain farts or
- 1:00:44computer farts so
- 1:00:46they're trying to solve these complexity problems and they have
- 1:00:50yet to do so and
- 1:00:52maybe I don't think they ever will I don't think they'll ever
- 1:00:56find a satisfactory
- 1:00:58SPEAKER_00: solution
- 1:01:02from you know for what they're looking for because I think like
- 1:01:05part of being in
- 1:01:06SPEAKER_00: the world part of being instantiated or part of like
- 1:01:10being in a complex situation is that you have to go down every
- 1:01:15path in order to see
- 1:01:16where it where it leads and you can't simulate that you can't
- 1:01:19just predict it but
- 1:01:22the the whole like Kolmogorov thing is really interesting to me
- 1:01:27I'm actually I've
- 1:01:28been I started a while ago writing a short story about the
- 1:01:32Kolmogorov complexity
- 1:01:35SPEAKER_00: and using it as kind of like a Voight-Kampff test
- 1:01:40Talking About AI And Humans And Stuff Like That But Yeah
- 1:01:44Complexity Theory
- 1:01:45SPEAKER_00: Complexity Science It's Interesting And I Don't Think There Are Any
- 1:01:50SPEAKER_00: Shortcuts when it comes to complexity.
- 1:01:52SPEAKER_00: I think complexity is resilient, at least for now anyway.
- 1:01:59SPEAKER_00: Maybe, who knows?
- 1:02:00Maybe in the future we'll have a different way of solving
- 1:02:05complexity problems.
- 1:02:06SPEAKER_00: But I think for now we just have to cope with complexity.
- 1:02:13SPEAKER_00: So the fifth point, self-improvement.
- 1:02:15Probably a truly intelligent machine will carry out activities
- 1:02:19which may be best
- 1:02:20SPEAKER_00: described as self-improvement.
- 1:02:22Some schemes for doing this have been proposed and are worth
- 1:02:26further study.
- 1:02:29SPEAKER_00: It seems likely that this question can be studied abstractly as well.
- 1:02:33SPEAKER_00: Yeah, self-improvement.
- 1:02:35SPEAKER_00: Uh, we improve ourselves over time.
- 1:02:37SPEAKER_00: I.J.
- 1:02:37SPEAKER_00: Goode talks about, you know, the intelligence explosion and what, uh,
- 1:02:44SPEAKER_00: Yeah, IJ Good.
- 1:02:45SPEAKER_00: Let's see here.
- 1:02:46SPEAKER_00: Speculations concerning the first ultra-intelligent machine.
- 1:02:49SPEAKER_00: Like, that's the whole thing.
- 1:02:52SPEAKER_00: The intelligence explosion self-improvement.
- 1:02:54SPEAKER_00: And yeah, I think that's a necessary thing for artificial intelligence.
- 1:02:57General Intelligence as we call it now but that AGI as we say
- 1:03:02today is what these
- 1:03:03guys were on you know 80 years ago it's just through marketing
- 1:03:10hype and through
- 1:03:10SPEAKER_00: bullshit boom and bust cycles and all that people now conflate artificial
- 1:03:17SPEAKER_00: intelligence with
- 1:03:19Your washing machine deciding that it's going to tell you three
- 1:03:23minutes until your
- 1:03:24SPEAKER_00: load is done.
- 1:03:25SPEAKER_00: And then,
- 1:03:26SPEAKER_00: you know,
- 1:03:26SPEAKER_00: you wait five minutes and then you make a cup of tea or,
- 1:03:29SPEAKER_00: you know,
- 1:03:30SPEAKER_00: you do something and then you're like,
- 1:03:31SPEAKER_00: oh shit,
- 1:03:31SPEAKER_00: the laundry's done.
- 1:03:32SPEAKER_00: And then you go back and you look and it says 25 more minutes.
- 1:03:35SPEAKER_00: You're like, what the fuck, bitch?
- 1:03:36SPEAKER_00: I gotta fault my laundry.
- 1:03:38SPEAKER_00: That's what people equate AI with nowadays and all the other bullshit.
- 1:03:44SPEAKER_00: But back when these guys kind of coined this term,
- 1:03:47They really meant something like what we mean when we say
- 1:03:52artificial general
- 1:03:53intelligence or artificial super intelligence today or
- 1:03:57artificial sentience or
- 1:03:58SPEAKER_00: whatever.
- 1:04:01SPEAKER_00: The sixth point, abstractions.
- 1:04:03A number of types of abstraction can be distinctly defined and
- 1:04:07several others less distinctly.
- 1:04:09A direct attempt to classify these and to describe machine
- 1:04:12methods of forming
- 1:04:13SPEAKER_00: abstractions from sensory and other data
- 1:04:16SPEAKER_00: Yeah,
- 1:04:21SPEAKER_00: I mean,
- 1:04:21SPEAKER_00: I think abstraction,
- 1:04:22SPEAKER_00: it's what we do when we think about things,
- 1:04:28SPEAKER_00: you know,
- 1:04:28SPEAKER_00: we can generalize,
- 1:04:30SPEAKER_00: we can analyze,
- 1:04:31SPEAKER_00: we can denature or de-world a notion or a concept or a thing.
- 1:04:39And we can from that try to distill some sort of essence or we
- 1:04:42can from that try to
- 1:04:43predict what it may or may not do in a given blah blah blah or
- 1:04:47all this stuff and
- 1:04:50SPEAKER_00: abstraction.
- 1:04:51SPEAKER_00: It's very important to human cognition.
- 1:04:55SPEAKER_00: I also think machines can probably do it.
- 1:05:01SPEAKER_00: Like I don't think there's now...
- 1:05:05SPEAKER_00: There is an element of negativity when it comes to abstraction.
- 1:05:10Because you have to de-world or de-nature the notion,
- 1:05:13the concept,
- 1:05:13the thing.
- 1:05:14SPEAKER_00: And you have to come up with a hypothetical.
- 1:05:17So it's like you're taking a thing and you're removing it from
- 1:05:19the world and you're
- 1:05:21SPEAKER_00: placing it in a void.
- 1:05:23SPEAKER_00: So I think negativity and even contradiction is important to abstraction.
- 1:05:33SPEAKER_00: But I also think
- 1:05:37Machines can replicate or simulate that activity just by doing
- 1:05:41something similar to
- 1:05:42the traveling salesman problem where you're just running it
- 1:05:45through hypotheticals
- 1:05:46hypothetical hypothetical hypothetical now is that the same
- 1:05:48thing as abstraction I
- 1:05:50don't know I don't think so but does it functionally achieve
- 1:05:53the same result yeah
- 1:05:56probably so I think they can simulate abstraction I don't think
- 1:06:03they can actually
- 1:06:04SPEAKER_00: perform abstraction
- 1:06:08But I also think this is probably another one of those
- 1:06:10instances of there's no
- 1:06:12SPEAKER_00: answer on this one yet because
- 1:06:16Yeah I don't even know rightly what they mean when they say
- 1:06:18abstraction I don't
- 1:06:19know proper like what do you mean abstraction proper do you
- 1:06:23mean math do you mean
- 1:06:26thinking about essences do you mean trying to predict things do
- 1:06:29you like what do
- 1:06:29you mean do you mean generalizing I don't know but yeah I don't
- 1:06:35think maybe
- 1:06:37colloquially like when you talk about abstraction you think of
- 1:06:41this quality of us
- 1:06:43SPEAKER_00: to kind of imagine
- 1:06:46Um And yeah I don't think machines can do that Properly I do
- 1:06:52think they can
- 1:06:52simulate it By doing the traveling salesman thing All like just
- 1:06:55boom boom boom boom
- 1:06:56SPEAKER_00: boom boom boom boom boom
- 1:06:59uh you know brawn over brains just doing it the hard way I
- 1:07:03think they can they can
- 1:07:04simulate it by doing it the hard way but they can't properly do
- 1:07:08it um because I
- 1:07:10think in order to properly do it you have to be able to hold
- 1:07:13negativity um and
- 1:07:16contradiction and you have to be able to think you know two
- 1:07:20contradictory things at
- 1:07:21SPEAKER_00: the same time and um
- 1:07:24SPEAKER_01: Randomness and Creativity
- 1:07:47A fairly attractive and yet clearly incomplete conjecture is
- 1:07:51that the difference
- 1:07:52SPEAKER_00: between creative thinking and unimaginative,
- 1:07:56SPEAKER_00: competent thinking lies in the injection of some randomness.
- 1:08:01SPEAKER_00: The randomness must be guided by intuition to be efficient.
- 1:08:04SPEAKER_00: In other words,
- 1:08:05the educated guess or the hunch include controlled randomness
- 1:08:08in otherwise orderly
- 1:08:10SPEAKER_00: thinking.
- 1:08:13SPEAKER_00: I don't know how I feel about that man I don't know I don't like the word
- 1:08:18creativity I don't know about randomness I don't know Intuition
- 1:08:28Yeah I do think
- 1:08:30there is something with like embodied knowledge I think there's
- 1:08:34something about the
- 1:08:34SPEAKER_00: life world There's something about kind of background Familiarity
- 1:08:43that uh allows us to be able to know things implicitly uh
- 1:08:50without maybe without oh
- 1:08:51shit my dog oh hold on he's about to go pee don't go pee in
- 1:08:54here hold on okay yeah
- 1:08:58I need to I need to fix my interface I'm gonna do that as soon
- 1:09:02as I'm done
- 1:09:02recording this uh actually as soon as I'm done getting this
- 1:09:06ready to upload I'm
- 1:09:07SPEAKER_00: gonna mess with my interface but uh
- 1:09:13Randomness,
- 1:09:14creativity,
- 1:09:15background,
- 1:09:15familiarity,
- 1:09:17intuition,
- 1:09:18umwelt,
- 1:09:19all these things.
- 1:09:20SPEAKER_00: I do think there's something to embodiedness.
- 1:09:24SPEAKER_00: I do think there's something in that kind of... Oh my gosh.
- 1:09:31SPEAKER_00: Fucking hold on.
- 1:09:32SPEAKER_00: Dogs.
- 1:09:34SPEAKER_00: Okay.
- 1:09:37SPEAKER_00: Okay.
- 1:09:38SPEAKER_00: I do think there's something wrapped up
- 1:09:50SPEAKER_00: Hans Moravec writes about having a body and how having a body pertains to
- 1:09:57intelligence and that's not included here in this Dartmouth
- 1:10:02proposal and that's not
- 1:10:03SPEAKER_00: included in in a lot of oddly a lot of
- 1:10:10I think John Searle does some interesting things with these
- 1:10:15ideas of embodiedness
- 1:10:17and trying to I think imminently critique functionalism and
- 1:10:23almost do like a
- 1:10:24SPEAKER_00: analytic phenomenology but yeah I think maybe Hans Moravec
- 1:10:30was kind of doing that but also kind of not because I think he
- 1:10:32was doing a
- 1:10:33functionalism I do think John Searle was doing something more
- 1:10:36than just another
- 1:10:37functionalism but I don't even know I don't think it comes down
- 1:10:43to necessarily just
- 1:10:44embodiment Merleau-Ponty I love Merleau-Ponty and I love having
- 1:10:50a well I don't love
- 1:10:51having a body I hate having a body actually but also my body is
- 1:10:54a condition of
- 1:10:54SPEAKER_00: possibility for me so you know take it or leave it
- 1:11:01Um But anyway Yeah I think it's I don't know Berkson talked
- 1:11:09about intuition And
- 1:11:10kind of Ilan Mittal And I don't know man There's just There's
- 1:11:16always this tendency
- 1:11:17to reduce Intelligence And even Things like randomness and
- 1:11:23creativity To like
- 1:11:24SPEAKER_00: reduce them down To some type of essence Or something easily identifiable
- 1:11:31um and I think they all get at aspects of it much like
- 1:11:34intelligence gets at an
- 1:11:35aspect of the human but it doesn't capture it doesn't contain
- 1:11:39it doesn't sum up uh
- 1:11:41it doesn't identify or communicate the human uh randomness and
- 1:11:46and creativity and
- 1:11:47background familiarity and intuition and life world and all
- 1:11:50this stuff that doesn't
- 1:11:52SPEAKER_00: embody this it doesn't
- 1:11:55Capture and Contain This Essence Of Similar To Abstraction As
- 1:11:59Well Like There's
- 1:12:00Just I Don't Know There's A Feature Of Like Being Here And
- 1:12:07Perceiving And Being In
- 1:12:08The World And Being Part Of The World But Also Being Like
- 1:12:11Radically Opposed To The
- 1:12:14World My Needs Are Not The Needs Of The World Like There's This
- 1:12:17Impasse There's
- 1:12:18SPEAKER_00: This Radical Breakdown
- 1:12:22and from that you get all these things creativity and
- 1:12:26randomness and the ability to
- 1:12:29try new things and do new things and the ability to fail and
- 1:12:32the ability to fail in
- 1:12:33SPEAKER_00: interesting ways and the ability to succeed and the ability to exceed in
- 1:12:37non-standard ways and all these things it's just like these
- 1:12:43guys are functionalists
- 1:12:46SPEAKER_00: and they're trying to break down intelligence and
- 1:12:50SPEAKER_00: G General Intelligence like they're trying to identify
- 1:12:55Characteristics and Features of Intelligence in a way that they
- 1:12:59can replicate it in a machine.
- 1:13:04SPEAKER_00: And I think that's,
- 1:13:06SPEAKER_00: again,
- 1:13:06going back to talking about modernity and previously we
- 1:13:11identified as some small
- 1:13:13SPEAKER_00: part of the divine or whatever.
- 1:13:16SPEAKER_00: And so we didn't have to think that we could replicate
- 1:13:21ourselves in machines we could think and we did think that we
- 1:13:24could encode our will
- 1:13:27we could replicate our agency we could replicate our desire or
- 1:13:32demand but we didn't
- 1:13:35identify with it like there was already this movement from the
- 1:13:38divine to the human
- 1:13:41and then on to the golem or on to the replicant or whatever
- 1:13:45here in modernity and
- 1:13:47SPEAKER_00: particularly with this paper but
- 1:13:50You know this paper is like the crystallization of a movement
- 1:13:52of a historical
- 1:13:53movement that's going on for a hundred years and we're you know
- 1:13:56almost a hundred
- 1:13:57years later we're still in this moment of like genuinely
- 1:14:02identifying with the
- 1:14:05SPEAKER_00: replication of will agency desire demand whatever
- 1:14:17SPEAKER_00: I don't know my thoughts on that.
- 1:14:19SPEAKER_00: I'm still thinking about that.
- 1:14:20SPEAKER_00: I'm still trying to get that all squared away so I can write about it.
- 1:14:29SPEAKER_00: There's a surplus lack.
- 1:14:31SPEAKER_00: There's a surplus negativity.
- 1:14:32There's a hole in the middle of the bagel and we have to
- 1:14:36include the hole.
- 1:14:37SPEAKER_00: The hole is part of the bagel.
- 1:14:39the hole is part of the donut you know you go to the donut
- 1:14:42store for me anyway when
- 1:14:43I go to the donut store and I say can I get a donut and they
- 1:14:46say what kind I say I
- 1:14:47don't know man surprise me I don't care dude and they give me
- 1:14:51one of those ones
- 1:14:52that's like a jelly filled donut I'm sad and I often I don't
- 1:14:57eat them I don't like
- 1:14:57jelly filled donuts or cream filled donuts um I like my donuts
- 1:15:01with holes in them
- 1:15:03SPEAKER_00: and if you say donut uh
- 1:15:07SPEAKER_00: It invokes images of a ring of dough,
- 1:15:11SPEAKER_00: maybe some frosting on it,
- 1:15:13SPEAKER_00: and a hole in the middle,
- 1:15:15SPEAKER_00: right?
- 1:15:15SPEAKER_00: That's what a donut is.
- 1:15:16SPEAKER_00: The hole makes up.
- 1:15:18SPEAKER_00: The hole is constitutive.
- 1:15:20SPEAKER_00: Now,
- 1:15:21SPEAKER_00: it is true that there are some donuts that don't have a hole,
- 1:15:24SPEAKER_00: and in fact,
- 1:15:24SPEAKER_00: they're filled with cream or jelly.
- 1:15:29SPEAKER_00: But yeah, the blah, blah, blah, blah, blah, whatever.
- 1:15:40There's all that but that is the those are the points they make
- 1:15:46we'll finish this
- 1:15:47section off and then we'll actually get into these guys kind of
- 1:15:51introducing
- 1:15:51themselves a little bit and talking and then we'll finish up so
- 1:15:55this shouldn't take
- 1:15:55too much longer in addition to the above collectively
- 1:16:00formulated problems for study
- 1:16:02we have asked the individuals taking part to describe what they
- 1:16:05will work on
- 1:16:06SPEAKER_00: statements by the four originators of the project are attached
- 1:16:10SPEAKER_00: We propose to organize the work of the group as follows.
- 1:16:14SPEAKER_00: Oh, this is where they talk about the money, isn't it?
- 1:16:16SPEAKER_01: Yeah.
- 1:16:17SPEAKER_00: Okay,
- 1:16:18potential participants will be sent copies of the proposal and
- 1:16:20asked if they would
- 1:16:21SPEAKER_00: like to work on the artificial intelligence problem of the group,
- 1:16:22SPEAKER_00: and if so,
- 1:16:23SPEAKER_00: what they would like to work on.
- 1:16:24The invitations will be made by the organizing committee on the
- 1:16:27basis of its
- 1:16:28estimate of the individual's potential contribution to the work
- 1:16:30of the group.
- 1:16:33So this is just saying these four guys kind of came up with
- 1:16:37this proposal and then
- 1:16:38SPEAKER_00: they're going to send out invitations to other people.
- 1:16:40SPEAKER_00: They introduce each other or they introduce themselves.
- 1:16:43SPEAKER_00: Claude Shannon,
- 1:16:43SPEAKER_00: mathematician,
- 1:16:44SPEAKER_00: Bell Telephone,
- 1:16:45SPEAKER_00: laboratories,
- 1:16:47SPEAKER_00: Marvin Minsky,
- 1:16:48SPEAKER_00: Harvard Junior Fellow in mathematics and neurology.
- 1:16:51Nathaniel Rochester Manager of Information Research IBM
- 1:16:54Corporation Poughkeepsie
- 1:16:56New York and Jay McCarthy John McCarthy Assistant Professor of
- 1:17:01Mathematics
- 1:17:02SPEAKER_00: Dartmouth College I believe McCarthy then went on to Stanford Stanford
- 1:17:08So they introduce who they are Then they say we need $1,200 a
- 1:17:15week And we need bus
- 1:17:19fare and rail fare and room and board and all this Because
- 1:17:24again this is a proposal
- 1:17:26They're saying if you come to this research project for two
- 1:17:30months We'll pay you
- 1:17:31SPEAKER_00: $12,000 a month or a week Or $700 if you're a grad student
- 1:17:37We will provide you with a secretary or a PA We'll pay your
- 1:17:41room and board We'll
- 1:17:43SPEAKER_00: give you bus fare,
- 1:17:44SPEAKER_00: rail fare,
- 1:17:45SPEAKER_00: blah,
- 1:17:45SPEAKER_00: blah,
- 1:17:45blah Proposal And then they're going to send out this proposal
- 1:17:50to the people at the
- 1:17:53bottom And then this is each guy gives a particular proposal So
- 1:18:00this is Claude
- 1:18:00SPEAKER_00: Shannon's proposal
- 1:18:02Claude Shannon,
- 1:18:04I would like to devote my research to one or both of the topics
- 1:18:07listed below.
- 1:18:09SPEAKER_00: While I hope to do so,
- 1:18:10it is possible that because of personal considerations I may
- 1:18:13not be able to attend
- 1:18:14SPEAKER_00: for the entire two months.
- 1:18:16SPEAKER_00: I nevertheless intend to be there for whatever time is possible.
- 1:18:20SPEAKER_00: The first point, topic.
- 1:18:23Application of Information Theory Concepts to Computing
- 1:18:26Machines and Brain Models A
- 1:18:28basic problem in information theory is that of transmitting
- 1:18:31information reliable
- 1:18:32SPEAKER_00: over a noisy channel.
- 1:18:37An analogous problem in computing machines is that of reliable
- 1:18:41computing using
- 1:18:43SPEAKER_00: unreliable elements.
- 1:18:45SPEAKER_00: This problem has been studied by von Neumann for Schaeffer Stroke...
- 1:18:50Excuse me by Von Neumann for Sheffer Stroke Elements and by
- 1:18:54Shannon and Moore for
- 1:18:56Relays but there are still many open questions and Moore Thomas
- 1:19:01Moore I believe is
- 1:19:01the guy that's the guy who came up with Moore's Law basically
- 1:19:05every 18 months cost
- 1:19:08halves or efficacy doubles basically you can look at it either
- 1:19:13way but there are
- 1:19:14SPEAKER_00: still many open questions the problem for several and lower bounds
- 1:19:21SPEAKER_00: The problem for several elements,
- 1:19:23SPEAKER_00: the development of concepts similar to channel capacity,
- 1:19:27the sharper analysis of upper and lower bounds on the required
- 1:19:30redundancy,
- 1:19:31SPEAKER_00: etc.,
- 1:19:32SPEAKER_00: are among the most important issues.
- 1:19:34Another question deals with the theory of information networks
- 1:19:37where information
- 1:19:38SPEAKER_00: flows in many closed loops,
- 1:19:40as contrasted with the simple one-way channel usually
- 1:19:43considered in communication
- 1:19:44SPEAKER_00: theory.
- 1:19:45SPEAKER_00: Questions of delay become very important in the closed-loop case,
- 1:19:49SPEAKER_00: and a whole new approach seems necessary.
- 1:19:51This would probably involve concepts such as partial entropies
- 1:19:55when a part of the
- 1:19:56SPEAKER_00: past history of a message ensemble is known.
- 1:20:04like ISPs today and packets and stuff like that like it's stuff
- 1:20:11is not sent in one
- 1:20:13SPEAKER_00: one contiguous like a message is not sent in a contiguous transformation
- 1:20:19Transmission like it's broken up into packets and kind of
- 1:20:22blasted and then
- 1:20:24reassembled at the point where it's received so that's
- 1:20:28interesting to see that
- 1:20:29going back and Claude Shannon of course the progenitor of
- 1:20:33information theory that
- 1:20:35kind of all information can be reduced down to bits of
- 1:20:38information and then also
- 1:20:41SPEAKER_00: this idea that
- 1:20:45Noise and Signal which shout out Michel Serre but with with
- 1:20:49Claude Shannon like
- 1:20:52there's this idea that there's always a threshold of noise and
- 1:20:55signal it's
- 1:20:57interesting Claude Shannon's very interesting very important I
- 1:21:00will hopefully be
- 1:21:01SPEAKER_00: doing more with him in the future
- 1:21:04And then of course Moore Moore's Law they're thinking about the
- 1:21:07upper and lower
- 1:21:08bounds of the like the hardware what can this do and how can we
- 1:21:13make this more
- 1:21:15efficacious more effective not just efficient but also
- 1:21:19effective like increase its
- 1:21:20SPEAKER_00: capability but also reduce cost the second topic
- 1:21:26SPEAKER_00: The matched environment brain model approached to automata.
- 1:21:30In general,
- 1:21:30a machine or animal can only adapt to or operate in a limited
- 1:21:35class of environments.
- 1:21:37Even the complex human brain first adapts to the simpler
- 1:21:40aspects of its environment
- 1:21:41SPEAKER_00: and gradually builds up to the more complex features.
- 1:21:45I propose to study the synthesis of brain models by the
- 1:21:48parallel deployment of a
- 1:21:49series of matched theoretical environments and corresponding
- 1:21:53brain models which
- 1:21:54SPEAKER_00: adapt to them.
- 1:21:55The emphasis here is on clarifying the environmental model and
- 1:21:59representing it as a
- 1:22:00SPEAKER_00: mathematical structure.
- 1:22:02SPEAKER_00: Often,
- 1:22:02SPEAKER_00: in discussing mechanized intelligence,
- 1:22:04we think of machines performing the most advanced human thought
- 1:22:09activities proving
- 1:22:10SPEAKER_00: theorems,
- 1:22:12SPEAKER_00: writing music or playing chess.
- 1:22:15I am proposing here to start at the simple and when the
- 1:22:18environment is neither
- 1:22:20SPEAKER_00: hostile,
- 1:22:21SPEAKER_00: merely indifferent,
- 1:22:22nor complex and to work up through a series of easy stages in
- 1:22:27the direction of
- 1:22:28SPEAKER_00: these advanced activities.
- 1:22:31SPEAKER_00: That's something that resembles the game of life.
- 1:22:35SPEAKER_00: Start with simple, simple rules and allow complexity to grow from there.
- 1:22:40SPEAKER_00: And yeah, I think that's an important
- 1:22:49I think that's an important step in this process of creating a
- 1:22:55machine intelligence
- 1:22:56SPEAKER_00: or a general machine that can accomplish tasks in general,
- 1:23:02allowing this kind of complexity to emerge rather than trying
- 1:23:05to encode the
- 1:23:07SPEAKER_00: complexity into it from the start.
- 1:23:10And if you look above me on the screen,
- 1:23:12if you're watching on YouTube,
- 1:23:13you will see
- 1:23:16SPEAKER_00: Conway's Game of Life being played.
- 1:23:19SPEAKER_00: A little looping video of Conway's Game of Life being played.
- 1:23:23SPEAKER_00: Complex structures and rules emerging out of very,
- 1:23:26SPEAKER_00: very,
- 1:23:27SPEAKER_00: very simple environments and very simple rules.
- 1:23:31SPEAKER_00: Very complex structures can emerge.
- 1:23:32Complex four-dimensional structures,
- 1:23:34counting time as the fourth dimension.
- 1:23:37SPEAKER_00: Although this is a two-dimensional representation, so it's anyway...
- 1:23:46SPEAKER_00: A proposal for research by Marvin Minsky.
- 1:23:52It is not difficult to design a machine which exhibits the
- 1:23:55following type of learning.
- 1:23:57The machine is provided with input and output channels and an
- 1:24:00internal means of
- 1:24:01providing varied output responses to inputs in such a way that
- 1:24:04the machine may be
- 1:24:06trained by a trial and error process to acquire one of a range
- 1:24:11of input-output
- 1:24:12SPEAKER_00: functions.
- 1:24:13SPEAKER_00: Such a machine,
- 1:24:14when placed in an appropriate environment and given a criterion
- 1:24:19of success or
- 1:24:20SPEAKER_00: failure,
- 1:24:21SPEAKER_00: can be trained to exhibit goal-seeking behavior.
- 1:24:25Unless the machine is provided with or is able to develop a way
- 1:24:29of abstracting
- 1:24:30SPEAKER_00: sensory material,
- 1:24:31it can progress through a complicated environment only through
- 1:24:34painfully slow steps
- 1:24:36SPEAKER_00: and,
- 1:24:36SPEAKER_00: in general,
- 1:24:37SPEAKER_00: will not reach a high level of behavior.
- 1:24:40um so yeah this is just kind of taking ideas like the
- 1:24:45perceptron um with very very
- 1:24:49very simple or even the game of life you know very very simple
- 1:24:53binary win lose you
- 1:24:55know yes no on off whatever um things and without some ability
- 1:25:01to abstract or
- 1:25:02SPEAKER_00: complexify itself it's just too slow um
- 1:25:08Now let the criterion of success be not merely the appearance
- 1:25:12of a desired activity
- 1:25:13SPEAKER_00: pattern at the output channel of the machine,
- 1:25:16but rather the performance of a given manipulation in a given
- 1:25:19environment.
- 1:25:21Then in certain ways the motor situation appears to be a dual
- 1:25:25of the sensory
- 1:25:26situation and progress can be reasonably fast only if the
- 1:25:30machine is equally
- 1:25:31capable of assembling an ensemble of motor abstractions
- 1:25:35relating its output
- 1:25:36SPEAKER_00: activity to changes in the environment.
- 1:25:39Such motor abstractions can be valuable only if they relate to
- 1:25:43changes in the
- 1:25:43environment which can be detected by the machine as changes in
- 1:25:47the sensory
- 1:25:47SPEAKER_00: situation.
- 1:25:48SPEAKER_00: In other words, if they are related
- 1:25:56I have been studying such systems for some time and feel that
- 1:26:01if a machine can be
- 1:26:02designed in which the sensory and motor abstractions as they
- 1:26:06are formed can be made
- 1:26:07SPEAKER_00: to satisfy certain relations,
- 1:26:10SPEAKER_00: a high order of behavior may result.
- 1:26:13These relations involve pairing motor abstractions with sensory
- 1:26:17abstractions in
- 1:26:18such a way as to produce new sensory situations representing
- 1:26:23the changes in the
- 1:26:24environment that might be expected if the corresponding motor
- 1:26:27act actually took
- 1:26:28SPEAKER_00: place.
- 1:26:29The important result that would be looked for would be that the
- 1:26:31machine would tend
- 1:26:32to build up within itself an abstract model of the environment
- 1:26:35in which it is
- 1:26:36SPEAKER_00: placed If it were given a problem,
- 1:26:38it could first explore solutions within the internal abstract
- 1:26:41model of the
- 1:26:41environment and then attempt external experiments Because of
- 1:26:45this preliminary
- 1:26:46SPEAKER_00: internal study,
- 1:26:47these external experiments would appear to be rather clever and
- 1:26:50the behavior would
- 1:26:51SPEAKER_00: have to be regarded as rather imaginative
- 1:26:54And yeah,
- 1:26:54that's so internal world modeling and also the ability to
- 1:26:56relate,
- 1:26:57um,
- 1:27:17Basically Embodiment So Marvin Minsky Here Is Also Doing
- 1:27:22Something With Like Having
- 1:27:23A Body But Not Not Just Having A Body Having A Body And The
- 1:27:27Ability To Recreate A
- 1:27:29Spatio-Temporal Model Of The Body And The World Within The Body
- 1:27:35Itself Or Within
- 1:27:36The Model Itself So The Model Can Model The Model Can Model
- 1:27:40Itself And The World
- 1:27:41SPEAKER_00: Around It That Being This Kind Of Like Essential
- 1:27:46necessary feature trait characteristic of of an intelligent
- 1:27:50agent or whatever.
- 1:27:52SPEAKER_00: And yeah, I mean, we we model the world around us.
- 1:27:55SPEAKER_00: There's a lot of interesting,
- 1:27:58SPEAKER_00: you know,
- 1:27:58SPEAKER_00: interesting stuff in brain science,
- 1:28:01SPEAKER_00: whether they actually,
- 1:28:03SPEAKER_00: you know,
- 1:28:04Can See That That We Have Models And Other Sciences As Well You
- 1:28:10Know Different
- 1:28:10Types Of Psychology Memory Science And Stuff Like That Like
- 1:28:13It's Pretty Evident
- 1:28:14That We Run A Simulation Of The World Inside Our Brain And
- 1:28:18Within That Simulation
- 1:28:19SPEAKER_00: We Exist So We Model Ourselves We Have A Model
- 1:28:24SPEAKER_00: in which there is a model of us.
- 1:28:26SPEAKER_00: And we can even,
- 1:28:27SPEAKER_00: you know,
- 1:28:27go on deeper levels of abstraction where in that modeled world
- 1:28:33where there's a
- 1:28:33SPEAKER_00: model of us,
- 1:28:35SPEAKER_00: that model is modeling a world where there's a model of us within it.
- 1:28:38SPEAKER_00: And we can, again, this Matryoshka architecture.
- 1:28:42This thing keeps coming back up where it's like it has to the
- 1:28:47agent the whatever
- 1:28:49SPEAKER_00: the creature the intelligent being whatever has to have the ability to
- 1:28:55conceptualize itself as a meta system and the world around it
- 1:29:01as an environment so
- 1:29:04SPEAKER_00: yeah systems and subjects right but but yeah
- 1:29:12On this functionalist reading They're never accounting for
- 1:29:16breakdowns They're never
- 1:29:17accounting for lack The thing about the model of the world I
- 1:29:22have in my mind It's
- 1:29:25SPEAKER_00: never
- 1:29:27SPEAKER_00: It's never like a functional representation.
- 1:29:31SPEAKER_00: Here in my office at home,
- 1:29:32SPEAKER_00: you know,
- 1:29:32SPEAKER_00: I have,
- 1:29:33SPEAKER_00: there's padding on the wall,
- 1:29:35acoustic padding behind me here with little cool lights in it
- 1:29:38and over there and
- 1:29:39SPEAKER_00: over there and over there.
- 1:29:40SPEAKER_00: I have a corkboard with some stuff,
- 1:29:42SPEAKER_00: some,
- 1:29:42SPEAKER_00: you know,
- 1:29:43SPEAKER_00: Polaroid photos and some things,
- 1:29:45SPEAKER_00: posters,
- 1:29:46SPEAKER_00: photographs I've taken.
- 1:29:47SPEAKER_00: I have a poster.
- 1:29:50SPEAKER_00: Like my model of the world, like on that part of wall that's over there,
- 1:29:57I've paid no attention to it and I'll probably get it wrong and
- 1:30:00if I close my eyes
- 1:30:01and I try to walk through it and I put my hand out to it I
- 1:30:03would not accurately
- 1:30:05locate it but this wall right here that's right behind camera B
- 1:30:09with Polaroids and
- 1:30:11SPEAKER_00: family tchotchkes and shit like that I would be more
- 1:30:16SPEAKER_00: Apt to be able to recollect,
- 1:30:19SPEAKER_00: to remember,
- 1:30:20SPEAKER_00: to imagine,
- 1:30:21SPEAKER_00: to model that wall because there's significance to it.
- 1:30:24SPEAKER_00: But there's even things I would get wrong.
- 1:30:26Even that photograph that I have that I look at all the time
- 1:30:29when I'm in here,
- 1:30:30SPEAKER_00: when I'm working,
- 1:30:31SPEAKER_00: even if I close my eyes and I were to try to recreate it in my mind,
- 1:30:33SPEAKER_00: there would be subtle ways in which I get it wrong.
- 1:30:37Because we're not just doing a functional representation of the
- 1:30:42world when we
- 1:30:43SPEAKER_00: recreate the world,
- 1:30:44SPEAKER_00: when we model the world,
- 1:30:45SPEAKER_00: when we imagine the world because things that are of significance to us,
- 1:30:48SPEAKER_00: things that have meaning,
- 1:30:51SPEAKER_00: things that bear meaning to us,
- 1:30:53SPEAKER_00: either consciously or unconsciously,
- 1:30:55SPEAKER_00: either explicitly or implicitly,
- 1:30:57SPEAKER_00: they create breakdowns,
- 1:30:58SPEAKER_00: they create glitches.
- 1:31:01And so already where they're trying to recreate this
- 1:31:04intelligent agent that models
- 1:31:05SPEAKER_00: the world,
- 1:31:06SPEAKER_00: well,
- 1:31:06SPEAKER_00: that would be a purely functional model.
- 1:31:09SPEAKER_00: But the type of modeling that we do as subjects,
- 1:31:12SPEAKER_00: it's a necessarily it's an inherently dysfunctional.
- 1:31:15SPEAKER_00: It's an inherently glitchy.
- 1:31:16It's an it's an inherently fucked up model because meaning
- 1:31:20comes into play.
- 1:31:21SPEAKER_00: So it's it's like a it's like playing a Bethesda game.
- 1:31:24SPEAKER_00: It's like playing Fallout New Vegas.
- 1:31:28SPEAKER_00: and all the glitches,
- 1:31:30SPEAKER_00: all the problems with Fallout New Vegas,
- 1:31:33SPEAKER_01: but still coming back to play it over and over again.
- 1:31:35And if you don't play video games,
- 1:31:38I try not to make reference to them all the time.
- 1:31:41And if you do play video games and you don't like Fallout New
- 1:31:43Vegas,
- 1:31:44then you're a monster.
- 1:31:46And if you do play video games and you've never played Fallout
- 1:31:48New Vegas,
- 1:31:50give it a shot.
- 1:31:52SPEAKER_00: It's good.
- 1:31:55there's some moments in the game where it's very sparse I don't
- 1:31:58think they did a
- 1:31:59good job because Obsidian was kind of rushed in development so
- 1:32:05there's a lot of
- 1:32:05open expanse that they could have filled in with cool content
- 1:32:10but it's also I don't
- 1:32:11know if you get into the role playing aspect you can kind of
- 1:32:13role play like you are
- 1:32:15you know this courier out in the desert and no one's there and
- 1:32:18you're just walking
- 1:32:19SPEAKER_00: and it's all empty and desolate and stuff but anyway
- 1:32:22SPEAKER_00: If you don't get it, Bethesda games are notoriously buggy.
- 1:32:26SPEAKER_00: But they're also great games.
- 1:32:27SPEAKER_00: Many people love them.
- 1:32:27SPEAKER_00: I love them.
- 1:32:28SPEAKER_00: They're fun to play.
- 1:32:29SPEAKER_00: We keep coming back to them.
- 1:32:30SPEAKER_00: That's the way our representation of the world works in our mind.
- 1:32:32SPEAKER_00: They're glitchy.
- 1:32:33SPEAKER_00: They're dysfunctional.
- 1:32:34They're not perfect one-to-one replications of the world
- 1:32:37because they're always
- 1:32:38SPEAKER_00: heavy with meaning.
- 1:32:39SPEAKER_00: They're always heavy with significance from our subjective position.
- 1:32:43SPEAKER_00: Whereas a mechanical model,
- 1:32:44SPEAKER_00: this machine,
- 1:32:45SPEAKER_00: this computer,
- 1:32:47would seek to do a purely functional recreation of the world
- 1:32:50because it has to know
- 1:32:52SPEAKER_00: Um...
- 1:32:53SPEAKER_00: Am I going to be able to move here?
- 1:32:55SPEAKER_00: Can I do this?
- 1:32:56SPEAKER_00: Can I do that?
- 1:32:56SPEAKER_00: Can I do this?
- 1:32:57SPEAKER_00: XYZ There is no meaning.
- 1:33:02SPEAKER_00: There's no interiority for this type of machinic replication.
- 1:33:08SPEAKER_00: So that's already even, you know, we have a model in our heads.
- 1:33:12SPEAKER_00: Yes, it's true.
- 1:33:13We do run a simulation on the world in our heads,
- 1:33:14but also that simulation is necessarily buggy.
- 1:33:18SPEAKER_00: Whereas for these machine simulations,
- 1:33:20SPEAKER_00: they try to make them accurate and complete and one-to-one.
- 1:33:24SPEAKER_00: So that's another anti-functionalist point.
- 1:33:28Proposal for Research by Nathaniel Rochester Originality in
- 1:33:32Machine Performance
- 1:33:36SPEAKER_00: In writing a program for an automatic calculator,
- 1:33:39one ordinarily provides the machine with a set of rules to
- 1:33:43cover each contingency
- 1:33:44SPEAKER_00: which may arise and confront the machine.
- 1:33:47One expects the machine to follow this set of rules slavishly
- 1:33:51and to exhibit no
- 1:33:52SPEAKER_00: originality or common sense.
- 1:33:54SPEAKER_00: Furthermore,
- 1:33:55one is annoyed only at himself when the machine gets confused
- 1:33:59because the rules he
- 1:33:59SPEAKER_00: has provided for the machine are slightly contradictory.
- 1:34:03SPEAKER_00: Finally,
- 1:34:04SPEAKER_00: in writing programs for machines,
- 1:34:06SPEAKER_00: one sometimes must go at problems in a very laborious manner,
- 1:34:11whereas if the machine had just a little intuition or could
- 1:34:14make reasonable
- 1:34:15SPEAKER_00: guesses,
- 1:34:16SPEAKER_00: the solution of the problem could be quite direct.
- 1:34:19This paper describes a conjecture as to how to make a machine
- 1:34:22behave in a somewhat
- 1:34:23SPEAKER_00: more sophisticated manner in the general area suggested above.
- 1:34:28The paper discusses a problem on which I have been working
- 1:34:31sporadically for about
- 1:34:32SPEAKER_00: five years,
- 1:34:33and which I wish to pursue further in the artificial
- 1:34:37intelligence project next
- 1:34:39SPEAKER_00: summer.
- 1:34:43there are people who get very very frustrated with computers
- 1:34:50and I think yeah I so
- 1:34:55back in the 50s when they were writing this to use a computer
- 1:34:58you had to like write
- 1:35:00your own code and like they used punched cards and shit like
- 1:35:04that it was like
- 1:35:04physical media and so I can only imagine how much more
- 1:35:09frustrating it would be
- 1:35:11for the average everyday person to try to use a computer
- 1:35:14because in order to do
- 1:35:15anything on a computer back in the back in these days you
- 1:35:18literally had to write
- 1:35:20every everything you wanted it to do you had to design every
- 1:35:23every program every
- 1:35:24SPEAKER_00: process you had to design it from the ground up
- 1:35:28SPEAKER_00: And it would be very,
- 1:35:29SPEAKER_00: very frustrating,
- 1:35:30SPEAKER_00: I'm sure,
- 1:35:31if you made one small mistake and it didn't work and you had to
- 1:35:33go back and find it
- 1:35:34SPEAKER_00: and blah,
- 1:35:35SPEAKER_00: blah,
- 1:35:35SPEAKER_00: blah,
- 1:35:36SPEAKER_00: blah,
- 1:35:36SPEAKER_00: blah.
- 1:35:36But I can only imagine the average everyday person who gets mad
- 1:35:39at their fucking
- 1:35:40phone because their phone does something or it's not doing what
- 1:35:42they want or
- 1:35:43SPEAKER_00: whatever.
- 1:35:43SPEAKER_00: They throw their fucking phone on the ground and it explodes or whatever.
- 1:35:47Or their computer is not doing the intended thing and so they
- 1:35:49get angry and they're
- 1:35:50SPEAKER_00: like,
- 1:35:51SPEAKER_00: fucking stupid computer,
- 1:35:52SPEAKER_00: why can't it just do X,
- 1:35:53SPEAKER_01: Y,
- 1:35:53SPEAKER_01: and Z?
- 1:35:54like I can only imagine that person going back in time to the
- 1:35:571950s and trying to
- 1:35:59use a computer but yeah he's I think anticipating where he's
- 1:36:06going he wants
- 1:36:08SPEAKER_00: something like Google Gemini or
- 1:36:14Apple Siri or whatever why can't you just anticipate what I
- 1:36:17want you to do well
- 1:36:18it's the thing like so now you're at this point where
- 1:36:21previously computation or
- 1:36:24computers or technology or whatever was about encoding your
- 1:36:27will your explicit will
- 1:36:31SPEAKER_00: on the universe
- 1:36:34or on the cosmos or on the fabric of time and space itself or
- 1:36:37whatever like you
- 1:36:38want something to be done and so you do it or you make a
- 1:36:41machine to do it I think
- 1:36:43SPEAKER_00: what Rochester here wants is a way to encode your implicit will
- 1:36:49or a way to have a machine that anticipates what you want it to
- 1:36:53do before you tell
- 1:36:54SPEAKER_00: it what you want it to do.
- 1:36:56SPEAKER_00: So you never have to run into those problems of the unconscious,
- 1:37:00SPEAKER_00: those problems of using language itself and those breakdowns that are
- 1:37:07Both responsible for and the source of and results of the rich
- 1:37:15human meaning,
- 1:37:16SPEAKER_00: subjectivity itself.
- 1:37:18SPEAKER_00: So Rochester wants to eliminate subjectivity.
- 1:37:21SPEAKER_00: He wants to eliminate lack.
- 1:37:24He wants to eliminate the gap between demand and ultimately the
- 1:37:29failure to satisfy the demand.
- 1:37:32He wants to eliminate desire entirely and he just wants a
- 1:37:34purely functional world.
- 1:37:38because we've created a world with predictive machines that can
- 1:37:41anticipate your every need.
- 1:37:43SPEAKER_00: It sounds strikingly contemporary when you put it that way.
- 1:37:48I think Android is releasing Android 17 and I've seen a little
- 1:37:54bit about it and
- 1:37:55SPEAKER_00: they're just trying to integrate Gemini into everything and it's like,
- 1:37:58SPEAKER_00: bitch,
- 1:37:58SPEAKER_00: I don't want Gemini on everything I do on my cell phone.
- 1:38:02I don't want Gemini in my text messages I don't want Gemini in
- 1:38:05my messenger chats I
- 1:38:07don't want Gemini on my camera I don't want Gemini in my
- 1:38:10fucking Google Docs when
- 1:38:12I'm fucking writing Gemini is already all over Google Docs and
- 1:38:16it's like you can't
- 1:38:17even just do a spell check anymore like if you go to do like a
- 1:38:21spell check now it's
- 1:38:22SPEAKER_00: like well you want me to rephrase that for you and I've used it
- 1:38:26and it's sometimes it's like oh that's pretty good and other
- 1:38:29times it's like no no
- 1:38:30no dude you you got exactly you got what I was trying to say
- 1:38:33exactly wrong but
- 1:38:36they're trying to create these like uncannily predictive
- 1:38:41machines that can't
- 1:38:42anticipate your every wish every desire I think some of the
- 1:38:44marketing is like it
- 1:38:46SPEAKER_00: shows a person walking in a city and they see a poster for a
- 1:38:51a fucking performance or whatever at a theater and they're like
- 1:38:54hey Google fucking
- 1:38:56book a ticket for me and my homie to go to that show and then
- 1:38:59also make dinner
- 1:39:00plans or whatever and Google's like sure thing boss I got it
- 1:39:03homie and it's like
- 1:39:06what the fuck dude I don't know like it's just it's the same
- 1:39:09thing as the people
- 1:39:09who when they get dressed in the morning they lay out clothes
- 1:39:13and then they like
- 1:39:14they're like hey chat GPT is this a cool outfit will this make
- 1:39:17me cool if I wear
- 1:39:18this outfit chat GPT and it's like dude just put the fucking
- 1:39:22clothes on dude or
- 1:39:24also don't go outside naked and see what happens you know I
- 1:39:28don't fucking know man
- 1:39:29SPEAKER_00: it's just it's so ridiculous
- 1:39:31It sucks ass dude But that's what Rochester Was asking for And
- 1:39:38he I don't think he
- 1:39:39was aware of it I think he was genuinely fucking pissed Because
- 1:39:41he had to sit and
- 1:39:42make punch cards Every time he wanted the computer to do
- 1:39:45something for him He had
- 1:39:46to write the code From the fucking ground up From scratch you
- 1:39:50know And he's like
- 1:39:51man it would be cool if we didn't have this But You know shout
- 1:39:56out Norbert Wiener
- 1:39:57SPEAKER_00: Hearkening to the Sorcerer's Apprentice or that old parable,
- 1:40:03SPEAKER_00: the old,
- 1:40:04SPEAKER_00: you know,
- 1:40:04SPEAKER_00: aphorism,
- 1:40:05SPEAKER_00: be careful what you wish for.
- 1:40:07I think Rochester was a little overzealous and he wasn't being
- 1:40:10careful what he wished for.
- 1:40:11SPEAKER_00: And I think...
- 1:40:12SPEAKER_00: The Process of Invention or Discovery
- 1:40:31Living in the environment of our culture provides us with
- 1:40:35procedures for solving many problems.
- 1:40:38SPEAKER_00: Just how these procedures work is not yet clear,
- 1:40:40but I shall discuss this aspect of the problem in terms of a
- 1:40:43model suggested by
- 1:40:45SPEAKER_00: Craig.
- 1:40:46He suggests that mental action consists basically of
- 1:40:50constructing little engines
- 1:40:51inside the brain which can simulate and thus predict
- 1:40:55abstractions relating to
- 1:40:56SPEAKER_00: environment.
- 1:40:57Thus,
- 1:40:58the solution of a problem which one already understands is done
- 1:41:01as follows.
- 1:41:04SPEAKER_00: Oh shit, hold on.
- 1:41:06SPEAKER_00: Okay, we're back.
- 1:41:07Thus,
- 1:41:08the solution of a problem which one already understands is done
- 1:41:13as follows.
- 1:41:15One,
- 1:41:15the environment provides data from which certain abstractions
- 1:41:18are formed.
- 1:41:20SPEAKER_00: Two,
- 1:41:21SPEAKER_00: the abstractions,
- 1:41:22SPEAKER_00: together with certain internal habits or drives,
- 1:41:25provide a definition of a problem in terms of desired condition
- 1:41:30to be achieved in
- 1:41:31SPEAKER_00: the future,
- 1:41:32SPEAKER_00: a goal,
- 1:41:34SPEAKER_00: a suggested action to solve the problem.
- 1:41:38SPEAKER_00: In a simulation,
- 1:41:39SPEAKER_00: excuse me,
- 1:41:40stimulation to arouse in the brain the engine which corresponds
- 1:41:44to this situation.
- 1:41:46SPEAKER_00: Three,
- 1:41:47then the engine operates to predict what this environmental
- 1:41:50situation and the
- 1:41:51SPEAKER_00: proposed reaction will lead to.
- 1:41:54SPEAKER_00: And four,
- 1:41:55SPEAKER_00: if the prediction corresponds to the goal,
- 1:41:57SPEAKER_00: the individual proceeds to act as indicated.
- 1:42:00The prediction will correspond to the goal if living in the
- 1:42:03environment of his
- 1:42:04SPEAKER_00: culture has provided the individual with the solution to the problem.
- 1:42:08SPEAKER_00: Regarding the individual as a stored program calculator,
- 1:42:11SPEAKER_00: the program contains rules to cover this particular contingency.
- 1:42:15SPEAKER_00: For a more complex situation, the rules might be more complicated.
- 1:42:19The rules might call for testing each of a set of possible
- 1:42:22actions to determine
- 1:42:23SPEAKER_00: which provides the solution.
- 1:42:26SPEAKER_00: A still more complex set of rules might provide for uncertainty about the
- 1:42:30SPEAKER_00: environment,
- 1:42:30SPEAKER_00: as,
- 1:42:31SPEAKER_00: for example,
- 1:42:32SPEAKER_00: in playing tic-tac-toe,
- 1:42:33SPEAKER_00: one must not only consider his next move,
- 1:42:35SPEAKER_00: but the various possible moves of the environment or his opponent.
- 1:42:40Now consider a problem for which no individual in the culture
- 1:42:43has a solution and
- 1:42:44SPEAKER_00: which has resisted efforts at solution.
- 1:42:47SPEAKER_00: This might be a typical current unsolved scientific problem.
- 1:42:51The individual might try to solve it and find that every
- 1:42:54reasonable action led to failure.
- 1:42:56SPEAKER_00: In other words,
- 1:42:57SPEAKER_00: the stored program contains rules for the solution of this problem,
- 1:43:00SPEAKER_00: but the rules are slightly wrong.
- 1:43:04SPEAKER_00: He could get such behavior by trying different things at random,
- 1:43:17SPEAKER_00: but such an approach would usually be too inefficient.
- 1:43:21SPEAKER_00: There are usually too many possible courses of action,
- 1:43:24SPEAKER_00: of which only a tiny fraction are acceptable.
- 1:43:27The individual needs a hunch,
- 1:43:29something unexpected but not altogether unreasonable.
- 1:43:39SPEAKER_00: Excuse me, not altogether reasonable.
- 1:43:41SPEAKER_00: Some problems,
- 1:43:42SPEAKER_00: often those which are fairly new and have not resisted much effort,
- 1:43:46SPEAKER_00: need just a little randomness.
- 1:43:48SPEAKER_00: Others,
- 1:43:49SPEAKER_00: often those which have long resisted solutions,
- 1:43:51SPEAKER_00: need a really bizarre deviation from traditional methods.
- 1:43:55A problem whose solution requires originality could yield to a
- 1:44:00method of isol...
- 1:44:03A problem whose solution requires originality could yield to a
- 1:44:06method of solution
- 1:44:07SPEAKER_00: which involved randomness.
- 1:44:09SPEAKER_00: In terms of Crake's model,
- 1:44:12the engine which should simulate the environment at first fails
- 1:44:15to simulate
- 1:44:16SPEAKER_00: correctly.
- 1:44:17SPEAKER_00: Therefore,
- 1:44:18it is necessary to try various modifications of the engine
- 1:44:21until one is found that
- 1:44:23SPEAKER_00: makes it do what is needed.
- 1:44:25SPEAKER_00: So far,
- 1:44:26the nearest practical approach using this method in machine's
- 1:44:28solution of problems
- 1:44:29SPEAKER_00: is an extension of the Monte Carlo method.
- 1:44:50SPEAKER_00: In the usual problem,
- 1:44:51SPEAKER_00: which is appropriate for Monte Carlo,
- 1:44:53there is a situation which is grossly misunderstood and which
- 1:44:56has too many possible
- 1:44:57factors and one is unable to decide which factors to ignore in
- 1:45:01working out
- 1:45:01SPEAKER_00: analytical solutions.
- 1:45:03So the mathematician has the machine making a few thousand
- 1:45:06random experiments.
- 1:45:07The results of these experiments provide a rough guess as to
- 1:45:10what the answer may be.
- 1:45:12The extension of the Monte Carlo method is to use these results
- 1:45:15as a guide to
- 1:45:16determine what to neglect in order to simplify the problem
- 1:45:19enough to obtain an
- 1:45:20SPEAKER_00: approximate analytical solution.
- 1:45:23SPEAKER_00: It might be asked
- 1:45:46If they used system they would require impossibly detailed
- 1:46:01communication.
- 1:46:02If they used one system maybe they would require impossibly
- 1:46:05detailed communication.
- 1:46:07For the individual maturing in competition with other
- 1:46:10individuals the requirements
- 1:46:11SPEAKER_00: of mixed strategy using game theory terminology favor randomness.
- 1:46:16SPEAKER_00: For the machine,
- 1:46:17randomness will probably be needed to overcome the
- 1:46:20short-sightedness and prejudices
- 1:46:21SPEAKER_00: of the programmer.
- 1:46:23SPEAKER_00: While the necessity for randomness has clearly not been proven,
- 1:46:26SPEAKER_00: there is much evidence in its favor.
- 1:46:34Okay so that whole the Monte Carlo thing the brute force thing
- 1:46:40introducing
- 1:46:40randomness again sure yeah but that's on a purely functional
- 1:46:47that's that's a purely
- 1:46:48functional world like when we're setting goals there's this
- 1:46:52element of I don't know
- 1:46:54embodied intelligence of background familiarity there's this
- 1:46:57element of intuition
- 1:46:58SPEAKER_00: of subjective intuition
- 1:47:01SPEAKER_00: That we make a decision unconsciously.
- 1:47:05SPEAKER_00: And that we've made it.
- 1:47:09SPEAKER_00: And there is no conscious, there may be no explicit justification.
- 1:47:13SPEAKER_00: So it may appear random.
- 1:47:18So doing this Monte Carlo thing where you have a whole bunch of
- 1:47:21things and you have
- 1:47:21SPEAKER_00: some,
- 1:47:22SPEAKER_00: pay attention to this,
- 1:47:22SPEAKER_00: don't pay attention to that,
- 1:47:23SPEAKER_00: pay attention to this,
- 1:47:24SPEAKER_00: don't pay attention to that,
- 1:47:25SPEAKER_00: and on and on and on and on.
- 1:47:27SPEAKER_00: But you're trying to solve it like a math problem.
- 1:47:29SPEAKER_00: And you're just concerned with pure positive function.
- 1:47:33When it comes to the human solving problems or accomplishing
- 1:47:37goals or just doing
- 1:47:38SPEAKER_00: things,
- 1:47:38SPEAKER_00: going about daily life,
- 1:47:42SPEAKER_00: it's not about randomness.
- 1:47:45SPEAKER_00: Now,
- 1:47:45SPEAKER_00: it is also true that we developed over eons,
- 1:47:50SPEAKER_00: hundreds of thousands of years,
- 1:47:52SPEAKER_00: millions of years even.
- 1:47:53SPEAKER_00: We
- 1:47:55Fucking Evolved You Know I Don't Know When Life Started Here On
- 1:47:57Earth Two Billion
- 1:47:59Fucking Years Ago I Don't Know But We've Been Evolving Over
- 1:48:03That Time And So We
- 1:48:04SPEAKER_00: Have Been This Assemblage Of Systems Complexifying Over Time And
- 1:48:10And we for whatever reason like our eyes the way our eyes work
- 1:48:13like if your eyes
- 1:48:15actually registered every little bit of information it would be
- 1:48:18overload it would
- 1:48:19be too much so the eye is a very refined system when it comes
- 1:48:23to what to ignore and
- 1:48:24what not to ignore and foveation you know when you're looking
- 1:48:28at something general
- 1:48:32things everything is out of focus but then you're looking at
- 1:48:34one thing and that one
- 1:48:35thing is in focus and what to pay attention to and this and
- 1:48:37that and blah blah blah
- 1:48:38SPEAKER_01: blah blah
- 1:48:41So at this mechanical functionalistic level at this bare
- 1:48:44material metabolic level
- 1:48:46yes you could replicate all these things and you could
- 1:48:49accomplish a machine that
- 1:48:51can go out to do a great many general tasks in general but when
- 1:48:56it comes to
- 1:48:56subjectivity there's also this element of creating trouble for
- 1:49:00ourselves and
- 1:49:02SPEAKER_00: there's also this element of desire I don't even know my own desire
- 1:49:10Nevertheless I seek the things I seek out and I achieve the
- 1:49:14things I achieve and I
- 1:49:15fail to achieve the things I fail to achieve and I fail to
- 1:49:19accomplish the things I
- 1:49:20fail to accomplish and there's all this stuff that's happening
- 1:49:23at the level of the
- 1:49:24unconscious it's all going on in the unconscious and that's not
- 1:49:30this world of
- 1:49:31function like that's not this positivistic world that's this
- 1:49:34purely negative world
- 1:49:37SPEAKER_00: So while I do think it's true you can make a very strong,
- 1:49:41SPEAKER_00: very robust general machine that can accomplish general tasks in general.
- 1:49:47That's still like the fact that we're identifying what we do as
- 1:49:52humans with that
- 1:49:53SPEAKER_00: type of general intelligence.
- 1:49:54SPEAKER_00: Like it's already reducing us down to positivism to functionalism.
- 1:50:00And it's like,
- 1:50:01yeah,
- 1:50:01homie,
- 1:50:02we're missing out on this rich world of meaning by doing that.
- 1:50:06SPEAKER_00: But let's finish.
- 1:50:08SPEAKER_00: The Machine with Randomness
- 1:50:35SPEAKER_00: Two approaches, however, appear to be reasonable.
- 1:50:46One of these is to find how the brain manages to do this sort
- 1:50:49of thing and copy it.
- 1:50:51The other is to take some class of real problems which require
- 1:50:54originality in their
- 1:50:55SPEAKER_00: solution and attempt to find a way to write a program to solve them.
- 1:51:00It is not clear which it would be quicker,
- 1:51:03nor how many years or generations it would take.
- 1:51:07Most of my effort along these lines has so far been on the
- 1:51:10former approach because
- 1:51:11I felt that it would be best to master all relevant scientific
- 1:51:14knowledge in order
- 1:51:15SPEAKER_00: to work on such a hard problem,
- 1:51:17and I already was quite aware of the current state of
- 1:51:19calculators and the art of
- 1:51:21SPEAKER_00: programming them.
- 1:51:23SPEAKER_00: The control mechanism of the brain
- 1:51:41Consider someone who is almost asleep,
- 1:51:43slightly drunk,
- 1:51:44or slightly feverish.
- 1:51:57SPEAKER_00: Or saying I gotta, you know, dot all the Ts and cross the Is.
- 1:52:02SPEAKER_00: That's a mistake.
- 1:52:03SPEAKER_00: I say it often.
- 1:52:05SPEAKER_00: You don't dot Ts and cross Is.
- 1:52:07SPEAKER_00: You cross Ts and dot Is.
- 1:52:09SPEAKER_00: So that, like, that's a mistake.
- 1:52:11SPEAKER_00: That's a failure.
- 1:52:12SPEAKER_00: That's an error.
- 1:52:13SPEAKER_00: But it still succeeds.
- 1:52:14SPEAKER_00: It still seems to make sense.
- 1:52:16SPEAKER_00: But it only makes sense due to this kind of background familiarity.
- 1:52:21SPEAKER_00: Language, right?
- 1:52:24SPEAKER_00: Anyway,
- 1:52:24perhaps the mechanism of the brain is such that a slight error
- 1:52:27in reasoning
- 1:52:29SPEAKER_00: introduces randomness in just the right way.
- 1:52:32Perhaps the mechanism that controls serial order in behavior
- 1:52:35guides the random
- 1:52:36SPEAKER_00: factor so as to improve the efficiency of imaginative processes over pure
- 1:52:40SPEAKER_00: randomness.
- 1:52:42SPEAKER_00: So that would be like an overseer that guides things,
- 1:52:46SPEAKER_00: that breaks things down into logical steps so that one error...
- 1:52:49So it would chop things down into semi-redundant remedial tasks
- 1:52:55or there would
- 1:52:56SPEAKER_00: always be some overlap so that if you fail in one thing,
- 1:52:59SPEAKER_00: it's never a total failure because you have overlap.
- 1:53:02SPEAKER_00: If you fail at two,
- 1:53:05it's not a total failure because one overlaps with two and
- 1:53:07three also overlaps with
- 1:53:08SPEAKER_00: two.
- 1:53:09SPEAKER_00: So this kind of overseer function...
- 1:53:14Some work has been done on simulating neural nets or on our
- 1:53:18automatic calculator.
- 1:53:20One purpose was to see if it would be thereby possible to
- 1:53:23introduce randomness in
- 1:53:24SPEAKER_00: an appropriate fashion.
- 1:53:26It seems to have turned out that there are too many unknown
- 1:53:29links between the
- 1:53:29activity of neurons and problem solving for this approach to
- 1:53:32work quite yet.
- 1:53:34The results have cast some light on the behavior of nets and
- 1:53:37neurons but have not
- 1:53:39SPEAKER_00: yielded a way to solve problems requiring originality.
- 1:53:42By next summer,
- 1:53:43this work will be finished and a final report will have been
- 1:53:44written.
- 1:54:09My program is to try next to write a program to solve problems
- 1:54:13which are members of
- 1:54:13some limited class of problems that require originality in
- 1:54:17their solution.
- 1:54:18It is too early to predict just what stage I will be in the
- 1:54:20next summer or just how
- 1:54:23SPEAKER_00: I will then define the immediate problem.
- 1:54:25However,
- 1:54:26the underlying problem which is described in this paper is what
- 1:54:29I intend to pursue.
- 1:54:30SPEAKER_00: In a single sentence,
- 1:54:31the problem is how can I make a machine which will exhibit
- 1:54:35originality in its
- 1:54:36SPEAKER_00: solution of problems?
- 1:54:38SPEAKER_00: And again, you know, taking the subject into account.
- 1:54:43Sure,
- 1:54:43there's randomness and creativity and originality and this and
- 1:54:46that.
- 1:54:47SPEAKER_00: But there's also the unconscious.
- 1:54:49And these guys,
- 1:54:49these functionalists,
- 1:54:50they don't account for the unconscious.
- 1:54:53So again,
- 1:54:54you can make a general machine that accomplishes general tasks
- 1:54:57in general.
- 1:54:58and it could be very robust and very awesome and very
- 1:55:00impressive but it would not
- 1:55:03it would not be of the same nature as the human intelligence
- 1:55:08the human subject or
- 1:55:09whatever and that's the thing not trying to you know pick apart
- 1:55:14Nathaniel Rochester
- 1:55:15or anything or pick on these guys these guys are super duper
- 1:55:18important but it's
- 1:55:21SPEAKER_00: kind of like the
- 1:55:23the background of today the environment that we live in today
- 1:55:26people people just
- 1:55:27reduce us like everyone's a fucking functionalist everyone in
- 1:55:30this conversation all
- 1:55:31SPEAKER_00: these AI conversations that are going on all these people are fucking
- 1:55:35functionalists and it's like hold on dude fucking wait a minute
- 1:55:38man fucking no dude
- 1:55:41SPEAKER_00: it's all about the breakdown it's all about the glitch it's all about the
- 1:55:44SPEAKER_00: dysfunction it's all about you know the implicit
- 1:55:50SPEAKER_00: We are doomed to be nothing but cogs,
- 1:55:52SPEAKER_00: nothing but functions,
- 1:55:53SPEAKER_00: operations,
- 1:55:53SPEAKER_00: nothing but machines,
- 1:55:54SPEAKER_00: completing general tasks in general.
- 1:55:59SPEAKER_01: Never mind your pesky
- 1:56:15Then we have this proposal for research by John McCarthy and
- 1:56:18then that's it dude
- 1:56:19SPEAKER_00: proposal for research by John McCarthy
- 1:56:43During next year and during the summer research project on
- 1:56:46artificial intelligence,
- 1:56:47SPEAKER_00: I propose to study the relation of language to intelligence.
- 1:56:51It seems clear that the direct application of trial and error
- 1:56:54methods to the
- 1:56:55relation between sensory data and motor activity will not lead
- 1:56:59to any very
- 1:57:01SPEAKER_00: complicated behavior.
- 1:57:03SPEAKER_00: Rather,
- 1:57:03it is necessary for the trial and error methods to be applied
- 1:57:06at a higher level of
- 1:57:08SPEAKER_00: abstraction.
- 1:57:09The human mind apparently uses language as its means of
- 1:57:12handling complicated phenomena.
- 1:57:14The trial and error processes at a higher level frequently take
- 1:57:17the form of
- 1:57:18SPEAKER_00: formulating conjectures and testing them.
- 1:57:20The English language has a number of properties which every
- 1:57:23formal language
- 1:57:24SPEAKER_00: described so far lacks.
- 1:57:27One,
- 1:57:27arguments in English supplemented by informal mathematics can
- 1:57:31be concise.
- 1:57:32SPEAKER_00: Two,
- 1:57:33English is universal in the sense that it can set up any other
- 1:57:36language within
- 1:57:37SPEAKER_00: English and then use that language where it is appropriate.
- 1:57:41SPEAKER_00: Three,
- 1:57:41the user of English can refer to himself in it and formulate
- 1:57:45statements regarding
- 1:57:46SPEAKER_00: his progress in solving the problem he is working on.
- 1:57:49SPEAKER_00: Four,
- 1:57:50SPEAKER_00: in addition to rules of proof,
- 1:57:52SPEAKER_00: English,
- 1:57:52SPEAKER_00: if completely formulated,
- 1:57:54SPEAKER_00: would have rules of conjecture.
- 1:57:56The logical languages so far formulated have either been
- 1:57:59instruction lists to make
- 1:58:01computers carry out calculations specified in advance or else
- 1:58:05formalization of
- 1:58:06parts of mathematics The latter have been constructed so as to
- 1:58:11be easily described
- 1:58:12SPEAKER_00: in informal mathematics to allow translation of statements from informal
- 1:58:16mathematics into the language and to make it easy to argue
- 1:58:21about whether proofs of
- 1:58:24SPEAKER_00: question mark,
- 1:58:24SPEAKER_00: question mark,
- 1:58:25SPEAKER_00: question mark
- 1:58:26SPEAKER_00: Nanceee Three Billion Nances
- 1:58:51SPEAKER_00: Well, shout out.
- 1:59:12Perl Shoutout C C++ Java and so on and so on all these computer
- 1:59:23languages that took
- 1:59:24cues from spoken natural languages English and stuff like that
- 1:59:30and I think now
- 1:59:31we're even seeing with what's being called vibe coding people
- 1:59:35are and I think
- 1:59:37SPEAKER_00: The overwhelming majority of people who are doing it, it's silly.
- 1:59:42SPEAKER_00: I think Rick Rubin did a thing where he was like,
- 1:59:46SPEAKER_00: yeah man,
- 1:59:46SPEAKER_00: I just eat mushrooms and vibe code and it's all good.
- 1:59:50SPEAKER_00: I fucking hate Rick Rubin, dude.
- 1:59:52SPEAKER_00: He's responsible for some great accomplishments,
- 1:59:56SPEAKER_00: some great music,
- 1:59:57SPEAKER_00: some great art,
- 1:59:58SPEAKER_00: but fuck Rick Rubin,
- 1:59:59SPEAKER_00: he sucks.
- 2:00:01SPEAKER_00: But I think,
- 2:00:02SPEAKER_00: you know,
- 2:00:02I think there are probably some people at high level who have
- 2:00:05like a high level
- 2:00:07SPEAKER_00: knowledge of computer coding who have,
- 2:00:09SPEAKER_00: you know,
- 2:00:10SPEAKER_00: played around with vibe coding or natural language coding.
- 2:00:14SPEAKER_00: And I don't know.
- 2:00:15SPEAKER_00: I just also think like I don't.
- 2:00:20SPEAKER_00: It's not just language.
- 2:00:23SPEAKER_00: Everything is language.
- 2:00:25SPEAKER_00: But but everything is also the contradiction between
- 2:00:31Matter and material and language itself so it's like everything
- 2:00:35is everything is
- 2:00:37language but also everything is material everything is real and
- 2:00:40it's there's a
- 2:00:41tension between the real and the symbolic and also the
- 2:00:45imaginary right so going to
- 2:00:48this thing where it's you can't just make like this
- 2:00:52structuralist argument and
- 2:00:54reduce everything down to function and operation like there has
- 2:00:57to there has to be
- 2:01:00SPEAKER_00: What do they fucking call it?
- 2:01:05When whiskey When you're letting whiskey sit In an oak barrel
- 2:01:10Some evaporates
- 2:01:13The Devil Share The Angel Share I don't fucking know I thought
- 2:01:15I was going to be
- 2:01:16clever and say something clever there but I failed to do so but
- 2:01:21yeah like there is
- 2:01:22this surplus lack there is this negative excess in everything
- 2:01:29and in all things
- 2:01:32where you can't just come up with a purely analytic purely
- 2:01:36functional purely
- 2:01:37SPEAKER_00: structural world and have it actually be
- 2:01:41A Living World With Human Subjects In It Now Again If You're
- 2:01:44Just Talking About
- 2:01:45Creating General Machines That Complete General Tasks In
- 2:01:49General Fucking By All
- 2:01:51Means Dude But People Have Taken This Thought And This Work And
- 2:02:00This Field Of
- 2:02:01Artificial Intelligence And They've Identified The Human With
- 2:02:05The Artificial
- 2:02:05SPEAKER_00: Intelligence And It's A Shame
- 2:02:09SPEAKER_00: Because that does just reduce us all down to utility.
- 2:02:12SPEAKER_00: And it reduces us all down to exchange value.
- 2:02:16SPEAKER_00: It reduces us all down to this essence.
- 2:02:21SPEAKER_00: The essence of value.
- 2:02:23SPEAKER_00: Which eradicates life and the living human underneath it all.
- 2:02:30SPEAKER_00: And it's a shame.
- 2:02:31SPEAKER_00: But that was it.
- 2:02:32SPEAKER_00: That was it.
- 2:02:36A proposal for the Dartmouth Summer Research Project on
- 2:02:40Artificial Intelligence.
- 2:02:42So this is where the term and the field of artificial
- 2:02:46intelligence originates from.
- 2:02:49SPEAKER_00: Keeping in mind it is kind of a concentration,
- 2:02:51SPEAKER_00: a crystallization,
- 2:02:52SPEAKER_00: a distillation of the previous hundred years of history.
- 2:02:58and it also has gone on to determine the following 80 years or
- 2:03:03so like it's very
- 2:03:04much the world we live in today um has not moved on from this
- 2:03:091955 proposal at
- 2:03:12SPEAKER_00: Dartmouth College so there's all that and um
- 2:03:19I will leave you to it I'm going to bounce I'm going to edit
- 2:03:21this I'm pretty sure
- 2:03:22the audio on this is going to be fucked up and I paused it a
- 2:03:24million times so it's
- 2:03:25going to be a motherfucker but I'm going to edit it I'm going
- 2:03:28to get it ready to
- 2:03:29upload and then I'm going to mess around with my interface and
- 2:03:34get my camera set up
- 2:03:36SPEAKER_00: and all that shit so don't be a stranger
- 2:03:40Bingo Bango Bongo Let me switch back here for those of you
- 2:03:44watching on YouTube and
- 2:03:45SPEAKER_00: Substack Don't be a stranger Do all the things,
- 2:03:48hit all the buttons Stay tuned Keep yourself apprised of what's
- 2:03:52going on Check out
- 2:03:54all the stuff Dangerous Maybe Check out everything going on at
- 2:03:57Philosophy Portal
- 2:03:58Lots of dope stuff going on And all the other homies and all
- 2:04:02the other road dogs
- 2:04:03SPEAKER_00: and all that stuff
- 2:04:05SPEAKER_00: Bingo Bango Bongo Don't Eat Yellow Snow We'll catch you on the next one.
- 2:04:09SPEAKER_00: Thanks for being here.
- 2:04:10SPEAKER_00: Thanks for spending some time with me.
- 2:04:15SPEAKER_00: And yep, San Diego means whale penis.
- 2:04:33SPEAKER_01: Nanceee Three Billion Nanceee Three Billion
- 2:04:52SPEAKER_01: I said it before and I'll say it again.
- 2:04:54SPEAKER_01: Life is pretty bad.
- 2:04:55SPEAKER_01: You don't stop and look around once in a while.
- 2:04:58SPEAKER_01: You could miss it.
- 2:04:59SPEAKER_01: It's that you don't have to laugh with these color problems in it.
- 2:05:02SPEAKER_01: That's good.
- 2:05:03SPEAKER_01: In other words, it's not enough to think that you're going to be a hit.
- 2:05:14I'm going to tell you that question has a great laugh.