Latest / General Magic & The Soul of Tech / Ep. 6 | Superintelligence, Fusion, and Faith | Marc Porat

Transcript

Ep. 6 | Superintelligence, Fusion, and Faith | Marc Porat · General Magic & The Soul of Tech

• 0:00
  SPEAKER_0: This is Mike Stern.
• 0:04
  SPEAKER_1: And I'm Tom "Hersh" Hershenson.
• 0:04
  SPEAKER_1: Welcome to General Magic & The Soul of Tech.
• 0:10
  SPEAKER_2: We believed we were building something that would change the world, and in some ways we did.
• 0:15
  SPEAKER_2: In others, we didn't understand what we were setting in motion.
• 0:19
  SPEAKER_2: The story of General Magic isn't over.
• 0:21
  SPEAKER_2: We're still living inside what it started.
• 0:27
  I'm joined by our very special guest,
• 0:29
  Marc Porat,
• 0:30
  the co-founder of General Magic and one of the thought leaders about
• 0:34
  where AI is taking us and where we may or may not want to go with it.
• 0:38
  You know,
• 0:39
  Marc Porat,
• 0:39
  I think, stands apart as someone who has lived in the future for a long time.
• 0:43
  SPEAKER_1: That was true with digital communications and the advent of the information age.
• 0:48
  SPEAKER_1: And it's true again with the dawning of the AI age.
• 0:52
  So it's a thrill to speak to such a knowledgeable and far sighted,
• 0:57
  visionary,
• 0:59
  He's proven himself again and again that he sees things of consequence far
• 1:04
  in advance of other people.
• 1:06
  In the Magic movie there's a segment where you talk about when you're at Aspen
• 1:10
  and you just found yourself in the future and you saw social mobile and
• 1:13
  the internet happen and it was real and you're in it.
• 1:16
  Have you had a similar kind of epiphany around what you've been describing as
• 1:21
  the upside of AI, the abundant side of what superintelligence can deliver?
• 1:26
  SPEAKER_0: But have you lived in that future in the same way?
• 1:28
  SPEAKER_3: It's happening just as we speak for the first time since then.
• 1:33
  I thought during the intervening years between General Magic and now
• 1:37
  that I saw some things having to do with climate.
• 1:39
  SPEAKER_3: The experience that you just described, Mike, at Aspen was was visceral.
• 1:42
  And I did not have that around climate,
• 1:44
  unfortunately, because that's that's a civilizational scale problem.
• 1:48
  But lately I'm working on on two projects and they are beginning
• 1:54
  to manifest exactly what you asked about.
• 1:56
  So I am like on an arc that could dissolve into illusion and
• 2:01
  then I think I have enough perspective to back off and say this isn't it either.
• 2:04
  But right now they're they are very strong candidates and is unbelievably energizing
• 2:08
  for reasons that both of you lived lived through.
• 2:10
  SPEAKER_3: We lived through them together, you know, so we know what it what that feels like.
• 2:13
  Yeah,
• 2:14
  although this time I think I'm a lot more aware of the potential dystopian outcomes
• 2:19
  of what's happening around us,
• 2:20
  in a way that I really wasn't,
• 2:22
  and I don't think any of us were,
• 2:23
  but that's something I'd like you to challenge, if you think so.
• 2:26
  None of us saw the Magic device that we were creating as a gateway to oppression and
• 2:31
  the destruction of of community and the psychographic targeting of people and all
• 2:35
  the rest of that.
• 2:37
  I think we're much more aware now of potential downsides about AI,
• 2:40
  but lack much vision about how to deliver the abundance version and forestall
• 2:45
  the dystopian version.
• 2:47
  So one of the things you were saying in your Detroit talk struck me
• 2:50
  as eminently challengeable about the abundance piece.
• 2:54
  You talked about having your digital twin will kind of wrangle all of the agents
• 3:00
  that you're able to hire from your family doctor who will be better
• 3:03
  than any human doctor could be to your family lawyer same and so on
• 3:08
  but they'll you know they'll be freely available 10 cents an hour was
• 3:11
  the metaphor you used.
• 3:13
  SPEAKER_3: Yeah.
• 3:13
  How do we get from pricing by token today which makes it unaffordable for just
• 3:18
  about everybody to that kind of democratic access?
• 3:22
  SPEAKER_0: Trace the path
• 3:23
  Yeah, so first of all,
• 3:25
  at at film festivals,
• 3:26
  for for the General Magic film,
• 3:29
  we ended up apologizing to the audience when the audience brought up that point.
• 3:32
  And I think that the answer in part is that we had not seen dystopian technology at
• 3:37
  the at the scale that social media foisted on us and and took an election away and
• 3:43
  so on, and all the things you said, which I agree with.
• 3:45
  SPEAKER_3: We had not seen that.
• 3:46
  And secondly,
• 3:47
  there is a there's a self-induced illusion that actually I do refer to in the in
• 3:51
  the film, that to do what we did, one has to, in some sense, suppress the negative.
• 3:57
  SPEAKER_3: You just have to.
• 3:58
  SPEAKER_3: At least when we were young and and foolish.
• 4:00
  SPEAKER_3: And that's what happened then.
• 4:02
  And the lesson since then very much corresponds,
• 4:05
  Mike, to your statement, which is do we now understand the the downside?
• 4:10
  SPEAKER_3: The downside, the dystopianism has been has been discussed by really learned people.
• 4:15
  I'm sure you've read the books and seen the podcasts about everything that
• 4:18
  that could end up in extinction.
• 4:20
  SPEAKER_3: Human extinction.
• 4:21
  SPEAKER_3: That's a very powerful word.
• 4:22
  SPEAKER_3: Extinction is a that's it.
• 4:24
  SPEAKER_3: Lights out.
• 4:25
  SPEAKER_3: And not- irresponsible people have said that.
• 4:28
  SPEAKER_3: Eliezer Yudkowsky is a New York Times bestseller, a hundred thousand books sold.
• 4:33
  SPEAKER_3: So someone is resonating to that.
• 4:35
  And on the other side,
• 4:36
  we have people like Peter Diamandis and his and his cohort that are
• 4:38
  that are just full of abundance all the time.
• 4:40
  SPEAKER_3: And every point in between, I think, is is is populated by someone, really smart.
• 4:45
  So what that leads us to is something that we have to be able to,
• 4:49
  number one...
• 4:50
  someone like me who does see the the possibilities,
• 4:53
  you cannot do that and be intellectually honest or emotionally honest
• 4:57
  without honoring the other side.
• 4:59
  SPEAKER_3: Right?
• 4:59
  And when one is young and and forceful and slightly arrogant,
• 5:03
  is to when an objection comes along is to steamroll it,
• 5:05
  just like absolutely shatter it.
• 5:08
  SPEAKER_3: That's not how we can afford to be in this last and greatest revolution that we're experiencing.
• 5:14
  We have to honor every downside scenario that's thrown on the table
• 5:19
  before we can be honestly optimistic about some of the the benefits.
• 5:24
  SPEAKER_3: Now, the challenge is symmetric.
• 5:26
  SPEAKER_3: The duality is symmetric.
• 5:28
  SPEAKER_3: That's what duality is about.
• 5:29
  SPEAKER_3: You got you have to be able to hold two things in your mind at the same time.
• 5:32
  SPEAKER_3: So people who thoughtfully throw difficult things on the table.
• 5:38
  That same person needs to,
• 5:39
  to be intellectually honest, to actually articulate the upside for themselves.
• 5:43
  SPEAKER_3: And and then go ahead and be dystopian and destructive.
• 5:46
  SPEAKER_3: That's fine.
• 5:47
  SPEAKER_3: And that's how good dialogue happens, in my view.
• 5:50
  SPEAKER_1: Well, I think that there are a couple points to bring up.
• 5:52
  SPEAKER_1: I mean, first of all, I I would suggest that it's actually three things.
• 5:58
  SPEAKER_1: It's not being on the horns of dilemma or two different sides.
• 6:01
  SPEAKER_1: There's a a third possibility, which is we don't know.
• 6:06
  And people make strong claims on one side and strong claims on the other,
• 6:11
  but there also needs to be a strong claim of unknowability or yeah no.
• 6:17
  SPEAKER_1: Second of all, I think there's a a point to be made about inflection points, right?
• 6:22
  SPEAKER_1: So a big question with any technology is when becomes too late.
• 6:27
  I think of nuclear weapons and the famous doomsday clock,
• 6:31
  and how many minutes to midnight are we?
• 6:33
  I'm wondering what indicators,
• 6:37
  what early warnings or early indicators of relief would you identify with respect to
• 6:45
  the adoption of AI?
• 6:46
  Yeah, are are we surprised that that given that we're in the early,
• 6:51
  earliest days of this most powerful revolution...
• 6:54
  so put those things together:
• 6:56
  twelve hundred and seventy some odd days since of the Chat GPT moment,
• 7:00
  which is when AI,
• 7:01
  which of course is much older than that, burst on the scene as a thing.
• 7:05
  Twelve hundred and seventy-six days in the context of a revolution of this force
• 7:09
  and kinetic energy is nothing.
• 7:11
  So for us to imagine all kinds of outcomes to your point,
• 7:14
  Tom, without really knowing, is exactly where we need to be.
• 7:17
  SPEAKER_3: I mean, well, not need to be, but where we are, because that's the nature of these things.
• 7:22
  SPEAKER_3: So j just a little bit of historical context, and then I'll address your point more directly.
• 7:26
  Let's say that the agricultural revolution,
• 7:28
  the industrial revolution,
• 7:30
  created the same kind of wealth, abundance, and problems, a duality.
• 7:33
  And let's say that this one is going to create only the same amount,
• 7:36
  which I think is going to be a orders of magnitude more than
• 7:38
  the Industrial Revolution.
• 7:40
  Well, that thing,
• 7:41
  the first one,
• 7:42
  the first revolution took hundreds of years,
• 7:44
  and one could say 2,000, but let's just say 500 years to roll out.
• 7:48
  SPEAKER_3: And so we had a chance to see it, think about it, adjust to it.
• 7:52
  SPEAKER_3: Bad things happened, we tried to remedy them or not.
• 7:55
  SPEAKER_3: Civilizations were built on it and then destroyed, and so on.
• 7:58
  SPEAKER_3: But that played out.
• 8:00
  SPEAKER_3: That's our history.
• 8:01
  SPEAKER_3: Industrial, same thing.
• 8:02
  SPEAKER_3: It but it played out in a hundred and pick a number, twenty-five years.
• 8:06
  And and that was the room that we had to understand the goods and the bads
• 8:10
  and what we could do to mitigate the bads and take advantage of the goods.
• 8:13
  SPEAKER_3: This is happening in ten years.
• 8:16
  SPEAKER_3: Humans don't understand the the the velocity of exponential impact.
• 8:21
  SPEAKER_3: What do I mean by 10 years?
• 8:22
  In 10 years,
• 8:23
  the questions we talk about on this podcast will be like how could we have well,
• 8:28
  first of all, we should have, could have known.
• 8:30
  SPEAKER_3: We should have could have done things about it.
• 8:32
  SPEAKER_3: And besides, the big ones that actually happened, we didn't even talk about.
• 8:36
  SPEAKER_3: How could we have missed it?
• 8:38
  SPEAKER_3: And that's where I see the future.
• 8:39
  And so the intellectual cop-out of what I just said is we don't have to talk
• 8:43
  about it, because by definition, we don't know.
• 8:45
  But the actual point that you and Mike both made is we have to know,
• 8:49
  we have to engage, and we have to take action.
• 8:51
  And some of those actions might have to be radical,
• 8:54
  because if Michael's point is you know is taken,
• 8:57
  it could destroy big parts of civilization that we actually want.
• 9:02
  SPEAKER_0: Yeah, no, I mean that that that is the question.
• 9:05
  And again,
• 9:05
  what I took away from your Detroit talk is the tacit assumptions you were making is
• 9:09
  that the there are market solutions to what to do now.
• 9:11
  SPEAKER_0: And
• 9:12
  SPEAKER_3: Okay.
• 9:13
  SPEAKER_0: and again, that's something that I don't agree with.
• 9:15
  I mean I think we're seeing a colossal market failure in the massive misallocation
• 9:20
  of capital to the AI arms race and
• 9:22
  SPEAKER_3: Yeah, for sure.
• 9:23
  SPEAKER_0: to the extraordinary concentration of the benefits right now of building LLMs.
• 9:29
  SPEAKER_0: You know, it's going to 5,000 people in the valley.
• 9:31
  When when SpaceX, OpenAI,
• 9:33
  and Anthropic finish their public offerings,
• 9:35
  you know,
• 9:35
  we're gonna have 90 new billionaires and four
• 9:38
  or five hundred centimillionaires buying houses in the Bay Area.
• 9:42
  SPEAKER_0: I mean, that's where the value's going.
• 9:44
  SPEAKER_0: It's not being socialized at all.
• 9:45
  SPEAKER_0: So who who acts and how?
• 9:46
  SPEAKER_0: Where are the levers and who pushes them?
• 9:48
  SPEAKER_0: And I really want to get your take on that, you know.
• 9:50
  SPEAKER_3: Yeah.
• 9:51
  SPEAKER_3: okay.
• 9:54
  So the creation of an elite aristocracy that is completely detached from the reality
• 9:59
  of the rest of the world is something that has happened over and over and
• 10:03
  over again.
• 10:04
  SPEAKER_3: Unfortunately it's happening again.
• 10:05
  SPEAKER_3: And again, to the point of speed.
• 10:07
  Speed should not be underestimated as something psychological
• 10:11
  that we actually are not trained to understand.
• 10:14
  SPEAKER_3: We've never understood that kind of velocity and speed.
• 10:17
  So yeah,
• 10:18
  so suddenly there's a creation of a of an uber class,
• 10:21
  as there was in France,
• 10:22
  as there was in England,
• 10:23
  as there was in America back back when, as there was in the Roman times back when.
• 10:27
  SPEAKER_3: So that part isn't new.
• 10:29
  SPEAKER_3: What's new is the is the is the is the violence of it, small V violence.
• 10:33
  Although it could become large V violence,
• 10:36
  but let's say right now small V violence is like but it is explosion.
• 10:39
  SPEAKER_3: It's an explosion of energy.
• 10:41
  Is capital misallocated to all this ridiculous chase
• 10:45
  that says how much capex can I invest faster than you can invest into chips
• 10:49
  that are obsolete or depreciating like crazy in eight months?
• 10:51
  SPEAKER_3: I mean that the economics of that to any economist would say, well, no, don't do that.
• 10:56
  You know,
• 10:57
  Demis Hassabis,
• 10:59
  one of my heroes,
• 11:00
  is not a radical woolly optimist because he's actually working
• 11:04
  on theses like let's eradicate all disease,
• 11:08
  because disease is an engineering problem.
• 11:11
  SPEAKER_3: That that's a very grounded scientific approach to an abundance thing.
• 11:16
  SPEAKER_3: And number two, science.
• 11:17
  I mean, science should probably get the kind of capex and kind of intellect that the
• 11:22
  the cognitive side is getting.
• 11:24
  SPEAKER_3: By which I mean language.
• 11:25
  Yeah, language is cognition and and
• 11:28
  the best case is Steve Jobs is up there right now watching this podcast saying,
• 11:32
  Well, it used to be bicycles for the mind, now it's SpaceX for the mind.
• 11:35
  SPEAKER_3: I mean he's like, this thing is rocking.
• 11:38
  SPEAKER_3: And why isn't Apple doing more?
• 11:40
  SPEAKER_3: What's wrong with Apple?
• 11:41
  So, and the third one is okay,
• 11:43
  which is physical and world AI, and it's beginning to get its its thing.
• 11:47
  And the reason I say it's okay because I don't have profound use cases around it,
• 11:52
  even though it will have civilizational impact.
• 11:55
  Both the first and the third will destroy jobs,
• 11:59
  absolutely, and create personal chaos for those who have been destroyed.
• 12:03
  Now, putting on our hats as macro historians and macroeconomists,
• 12:07
  in the long run,
• 12:09
  jobs will find and will re-equilibrate to the new challenges or opportunities
• 12:13
  that these technological revolutions present.
• 12:17
  SPEAKER_3: People who have just lost their job do not live in the long run.
• 12:20
  SPEAKER_3: They live in, shit.
• 12:23
  I've actually spoken in these superintelligence keynotes to young people,
• 12:27
  23 years old,
• 12:27
  who actually come to me,
• 12:29
  you know, at the end crying about their existential blackness.
• 12:33
  But also,
• 12:34
  the last talk I gave was in Luxembourg,
• 12:36
  the president of one of the largest insurance companies in the world,
• 12:40
  said that about his own son, terrified, and then said that about himself.
• 12:45
  SPEAKER_3: This is the elite aristocracy head of a gigantic flywheel of a business called insurance.
• 12:52
  SPEAKER_3: President.
• 12:53
  SPEAKER_3: And a young president, not you know, not like retired.
• 12:56
  SPEAKER_3: So I think people who are vulnerable deserve to be angry and frightened.
• 13:04
  And at the same time,
• 13:05
  they also need to engage because there's a way out for those people
• 13:09
  who actually engage and want a way out.
• 13:14
  SPEAKER_3: I can describe that in slightly idealistic terms if you want.
• 13:17
  SPEAKER_0: Yeah, no.
• 13:17
  SPEAKER_0: Again, we're always looking for answers.
• 13:21
  SPEAKER_0: Or you know, they could they could be temporary, they could be, partial.
• 13:24
  SPEAKER_0: But yeah, I mean, again, where are the levers and who pushes them?
• 13:27
  SPEAKER_0: Both to mitigate the deleterious impacts and to amplify the benefits.
• 13:32
  SPEAKER_0: Because, you know, I'm completely with you on healthcare.
• 13:34
  SPEAKER_0: Yeah AI is going to revolutionize it in the best possible way.
• 13:38
  SPEAKER_0: If we can get to the point where we socialize.
• 13:40
  SPEAKER_0: its benefits.
• 13:40
  SPEAKER_0: And y so how do but how do we get there?
• 13:41
  SPEAKER_0: You know, that that that
• 13:42
  SPEAKER_3: No, okay.
• 13:42
  SPEAKER_3: So good.
• 13:43
  SPEAKER_3: So now Okay, so now
• 13:46
  SPEAKER_0: you know, because we're not doing that now.
• 13:47
  SPEAKER_0: We're not socializing its benefits, right?
• 13:49
  So the reason why why I think I might see something in the fog bank that begins
• 13:54
  to resemble civilizational scale impact is this project I'm doing,
• 13:58
  which I'm happy to tell you about called Alexandria Federation.
• 14:01
  SPEAKER_3: Alexandria as in "the Great Library of."
• 14:04
  SPEAKER_3: And in that it addresses exactly the point you said.
• 14:06
  SPEAKER_3: How do we democratize and make it available to everyone?
• 14:10
  SPEAKER_3: Not as a central decision by Sam Altman and and Elon Musk, but because of something else.
• 14:16
  SPEAKER_3: And I'll tell you about that right now.
• 14:17
  So in the Alexandria Federation idea,
• 14:20
  one notices that there are now 50 countries around the world who are objecting to
• 14:24
  the fact that the large language model that they use is 90% trained
• 14:28
  on American English, curated perspective period.
• 14:33
  SPEAKER_3: Why?
• 14:33
  SPEAKER_3: Because there are 18 trillion, you know, the parameters already just in English.
• 14:37
  SPEAKER_3: And and why do we need to go further?
• 14:39
  SPEAKER_3: So you pick a country, any country that has a great civilization, pick Mexico.
• 14:43
  SPEAKER_3: Mexico is at 0.4%, maybe, of a frontier model.
• 14:48
  SPEAKER_3: It absolutely ignores that civilization.
• 14:52
  There are other civilizations,
• 14:53
  I have a complete list,
• 14:54
  and countries that are point...
• 14:56
  pick a number of zeros like two, three, four, five, and that's it.
• 15:01
  So that's the outrage that I saw when I began to to think about that is the culture,
• 15:06
  the wisdom,
• 15:07
  the languages,
• 15:08
  the nuance,
• 15:09
  the learned experience,
• 15:10
  all of that from the rest of the world...
• 15:12
  is: Thank you very much, actually; we here in Silicon Valley, we don't really care.
• 15:17
  SPEAKER_3: It's not relevant because we're making trillions of dollars just doing the English thing.
• 15:21
  So there are 50 countries that have now started
• 15:23
  to do their own they call them "sovereign AI," but but but
• 15:27
  where they are collecting their corpora of their culture and the dominant languages
• 15:31
  and the dialects of their country, and they're building a stack.
• 15:34
  Unfortunately,
• 15:35
  none of those talk to each other in any technical way that could be called
• 15:39
  the emergence of a global OS for intelligence.
• 15:42
  SPEAKER_3: Step in Alexandria and says, hey.
• 15:44
  We're gonna create a fund,
• 15:45
  and we're gonna walk into as many countries as possible and hand you
• 15:49
  a very big check to federate so that your your language,
• 15:54
  dialect, culture is on an equal
• 15:56
  SPEAKER_0: Right.
• 15:56
  SPEAKER_3: basis with everyone else if they choose to to participate in.
• 15:59
  SPEAKER_3: So that's one.
• 16:00
  SPEAKER_3: Second, distribution.
• 16:02
  SPEAKER_3: How are you gonna do this thing?
• 16:03
  SPEAKER_3: Is this an idealistic piece of nonsense?
• 16:05
  SPEAKER_3: No.
• 16:06
  SPEAKER_3: Distribution is ironically the cell phone.
• 16:10
  SPEAKER_3: The smartphone, the very thing we worked on.
• 16:13
  SPEAKER_3: Why?
• 16:14
  SPEAKER_3: Because AI is Chat GPT on on an Android or whatever it is.
• 16:19
  And if you go in Alibaba right now and look for a phone,
• 16:24
  a cheap phone, you can find one equivalent to an iPhone 6 or 7 for $46.
• 16:30
  So that is cheap enough that that we can get down not to the base of the pyramid,
• 16:35
  but we can get down like two, three levels.
• 16:37
  SPEAKER_3: It's cheap enough that it it it's a global and what are you holding in your hand with for $46?
• 16:42
  SPEAKER_3: Supercomputer.
• 16:43
  SPEAKER_3: Do you need Wi-Fi to go access the global brain or any brain?
• 16:47
  SPEAKER_3: Yeah, no.
• 16:48
  SPEAKER_3: You need you need sort of, you know, mediocre cell service.
• 16:50
  Which as we know exists in in many of those countries because they leapfrogged over
• 16:54
  the the landlines.
• 16:56
  SPEAKER_3: So we have the makings of two things here if you put them together.
• 16:59
  SPEAKER_3: Thing one is the emergence of culturally authentic grounded proud of a culture.
• 17:07
  And secondly a distribution that is not guided by governments or by do good
• 17:12
  or committees, it just is.
• 17:13
  SPEAKER_3: That's the element of beginning to democratize it.
• 17:16
  SPEAKER_3: Now there's two two missing elements I should talk about.
• 17:19
  SPEAKER_3: One you'll accept, one you'll like, holy crap, this is ridiculous.
• 17:23
  SPEAKER_3: Okay, so the first one is the tech stack creates a Me.
• 17:26
  SPEAKER_3: So that anybody who wants a Me, which is not an obvious thing to know how to
• 17:30
  SPEAKER_0: Mm.
• 17:30
  do, it took me a couple of years to actually refine it to the point
• 17:33
  where like my Me is like amazing.
• 17:35
  SPEAKER_3: I'm living in AGI, for sure.
• 17:37
  SPEAKER_3: So AGI is not a function of the tech, necessarily, it's also function of the user.
• 17:43
  SPEAKER_3: So, you know, so I've got all the R & D, and the motive to to go create a perfect me.
• 17:48
  SPEAKER_3: But the rest of the world doesn't.
• 17:49
  So when you get one of these things burnt into the ROM of the phone for $46,
• 17:54
  the first thing you say, Would you like a digital twin?
• 17:56
  SPEAKER_3: A Me?
• 17:57
  SPEAKER_3: And yeah, that's it.
• 18:00
  And a lot of things happen in background,
• 18:02
  and now you can begin to to reach out and get the the world's best,
• 18:06
  teacher for that thing you want to know right now on Sunday afternoon,
• 18:09
  that that knows how you think and how you feel and how you learn.
• 18:13
  SPEAKER_3: You can do that because you have this thing.
• 18:15
  SPEAKER_3: That's a wonderful thing.
• 18:16
  SPEAKER_3: So Mia, we call it Mia.
• 18:18
  SPEAKER_3: It's the kind of the global, universal, democratized, cost you nothing, digital Me.
• 18:23
  SPEAKER_3: The second layer is Rosetta, as in the stone.
• 18:28
  Imagine if we had a modern AI,
• 18:30
  the the the union,
• 18:31
  the collection of all the civilizations,
• 18:34
  past and current,
• 18:35
  particularly current,
• 18:36
  and all the languages and dialects, accessible and available to anyone.
• 18:40
  So you and your dialect in southern Mexico,
• 18:43
  who wants to manage some market supplies something or other or wants to learn from
• 18:47
  a great teacher that doesn't live in their country, can do so.
• 18:50
  SPEAKER_3: Can do so in their dialect, in their cultural frame, in the nuance, because of this this thing.
• 18:56
  So that to me is is pretty far over on that whoopy to you know somewhat optimistic
• 19:02
  and utopian view.
• 19:03
  SPEAKER_3: It's a very hard project.
• 19:05
  SPEAKER_3: Now the one you're gonna absolutely say, Marc, you just stop what you're doing.
• 19:08
  SPEAKER_3: It's a complete waste time.
• 19:10
  SPEAKER_3: The one I also want to do is Contienza.
• 19:13
  SPEAKER_3: What is Contienza?
• 19:14
  SPEAKER_3: When we distribute this open source stack, it's open source.
• 19:19
  SPEAKER_3: And anybody can use it and and whatever they do comes back to it.
• 19:22
  SPEAKER_3: So kind of like, you know, global
• 19:23
  SPEAKER_0: Yeah.
• 19:24
  SPEAKER_3: AI hugging face, you know, GitHub thing.
• 19:26
  SPEAKER_3: So what's Contienza?
• 19:28
  When you have a lot of cultures with their corpora,
• 19:32
  their literature,
• 19:32
  their everything,
• 19:34
  in a kind of a wiki kind of model,
• 19:36
  so people can actually contribute and complain,
• 19:39
  and you know, so everyone is an editor.
• 19:41
  SPEAKER_3: Everyone is an editor that signs up to this, everyone.
• 19:45
  SPEAKER_3: So you get a cacophony of opinions.
• 19:48
  And you get you get people on one side who are screaming and yelling about you know
• 19:51
  about something,
• 19:52
  and on the other side, and they hate each other, they want to kill each other, etc.
• 19:55
  But but theoretically,
• 19:57
  in the center is something that I'm idealistically calling kind of the union or
• 20:02
  the natural consensus of what people who have
• 20:04
  who naturally hate each other would still agree to.
• 20:07
  SPEAKER_3: Would still agree to.
• 20:08
  SPEAKER_3: And I don't have a right to have a preconception about that.
• 20:11
  SPEAKER_3: I don't.
• 20:12
  That's the same thing as gathering a committee of wise people and saying we now have
• 20:16
  to do the the the you know the Bill of Rights and the first principles of AI.
• 20:19
  SPEAKER_3: That's a ridiculous.
• 20:20
  SPEAKER_0: Hey, it goes back to Plato.
• 20:21
  SPEAKER_0: I mean the problem goes back to Plato, right?
• 20:23
  SPEAKER_0: Y y y y you know, the Guardians, right?
• 20:25
  SPEAKER_0: I mean, yeah.
• 20:26
  SPEAKER_0: So who are they?
• 20:28
  SPEAKER_0: So
• 20:29
  SPEAKER_3: We we should not want guardians.
• 20:31
  I mean, it's fine,
• 20:32
  we can throw some in,
• 20:33
  but we should not to to do the fundamental thing that you're really driving at.
• 20:37
  How does AI find its its keel,
• 20:40
  its rudder,
• 20:41
  its its North Star,
• 20:41
  its direction,
• 20:43
  decoupled from the elites,
• 20:45
  the very elites,
• 20:46
  that are you know we will regulate or trust us,
• 20:49
  or we do want regulation, but as long as it's not real regulation?
• 20:53
  SPEAKER_3: That's the charade that you're complaining about.
• 20:55
  So Alexandria is an attempt to actually leapfrog that conceptually with a vision
• 20:59
  that says:
• 21:00
  w wh whatever meek and modest person has a need of any sort,
• 21:05
  they can reach through their supercomputer and and address that need in their frame
• 21:10
  and culture.
• 21:11
  Any anthropologist would tell you,
• 21:12
  if you're not talking to someone in their language,
• 21:15
  then you're not really communicating.
• 21:16
  SPEAKER_3: Yeah.
• 21:17
  SPEAKER_0: Okay.
• 21:18
  SPEAKER_0: But again, this goes who pays for it?
• 21:21
  SPEAKER_3: Yeah, okay.
• 21:21
  SPEAKER_3: the business model, you know, in some sense came early for Alexandria.
• 21:25
  SPEAKER_3: I'll tell you several aspects of it.
• 21:27
  One, So health health and education in any country,
• 21:31
  particularly ones that are socialized or have a national health service,
• 21:33
  take Mexico, for example, an enormous amount of waste of money with poor outcome.
• 21:40
  SPEAKER_3: And so part of the business model is to divert that into hyper personalized AI.
• 21:45
  SPEAKER_3: So someone is is well educated having a dialogue with their concierge, you know, doctor.
• 21:51
  Now, when I spoke with the Minister of Health of the country,
• 21:54
  and he said, Well, you know, there are type one and type two errors.
• 21:57
  SPEAKER_3: You know, people come rushing into emergency when they shouldn't.
• 21:59
  SPEAKER_3: You know, it's not a heart attack, it's the big greasy meal you just ate.
• 22:03
  They come rushing in to to over medicate,
• 22:05
  over surgery,
• 22:06
  and that is very expensive for for any health service,
• 22:09
  which is a big part of GDP, because it has lousy outcomes.
• 22:13
  SPEAKER_3: Type one error, don't come in.
• 22:15
  SPEAKER_3: Type two error.
• 22:16
  SPEAKER_3: They don't come in when they should.
• 22:17
  SPEAKER_3: So prenatal, postnatal, obesity, drug addiction, they don't come in.
• 22:22
  SPEAKER_3: And by the time you get them, in an ambulance, they cost an arm and a leg.
• 22:25
  SPEAKER_3: So where's this going in the business model question?
• 22:28
  I asked this Minister of Health,
• 22:30
  So is it worth 50 cents per citizen to to basically divert that
• 22:35
  into running this business, this Alexander business.
• 22:38
  SPEAKER_3: Because we think it has a 5x ROI to the to the country, to the ministry.
• 22:43
  SPEAKER_3: He'd been studying AI and medicine, really smart guy, had a little AI team with him.
• 22:48
  SPEAKER_3: He said, No, no, no, no, it's not 5x, no, no, no.
• 22:50
  SPEAKER_3: It's more like 10 or 20x.
• 22:52
  SPEAKER_3: So what does that mean?
• 22:53
  If 50 cents a month is good enough and you have 30 million people on it,
• 22:58
  that's $15 million a month at 80%, you know, or 90% margin.
• 23:02
  SPEAKER_3: That's a business.
• 23:03
  SPEAKER_3: Give you another one.
• 23:06
  So imagine talking to Samsung and saying, Hey,
• 23:09
  we're here to to have you burn this app into the ROM and to which they say
• 23:15
  to what advantage?
• 23:16
  SPEAKER_3: What's in it for us?
• 23:17
  To which the answer is imagine someone in country goes into a telco,
• 23:22
  and there are all these phones,
• 23:24
  and one of them speaks and thinks your language and culture.
• 23:28
  SPEAKER_3: You're more inclined to pick that one.
• 23:30
  SPEAKER_3: Point one.
• 23:30
  SPEAKER_3: So now there's a moat for a Samsung, or at least a competitive advantage for Samsung.
• 23:35
  SPEAKER_3: Thing two.
• 23:36
  All of these smartphone people want you to migrate to the newest latest version,
• 23:40
  even though the newest latest version is trivially more interesting than
• 23:43
  the last one.
• 23:44
  But if you walk into that same telco and you see your brand,
• 23:48
  you pick a Samsung or whatever it is,
• 23:50
  five models,
• 23:52
  and the newest the new one is actually thinks your language and culture,
• 23:56
  and the other ones don't think it.
• 23:58
  And one of them is,
• 23:59
  is a Me out of the box with medical,
• 24:01
  education and other benefits,
• 24:03
  you're more inclined to not use the five year old model.
• 24:06
  And any incremental migration up from last year's or five year ago model
• 24:10
  to this one is massively profitable to any smartphone manufacturer.
• 24:15
  SPEAKER_3: Those two things I just said is a business model.
• 24:17
  SPEAKER_3: Because you then say, I want rent for that competitive benefit.
• 24:20
  SPEAKER_1: Well, I just want to interject for a second.
• 24:22
  SPEAKER_1: I mean, you can think about different ways of funding, right?
• 24:25
  SPEAKER_1: And those are gonna probably tend towards different downsides.
• 24:30
  We can think of venture funding,
• 24:32
  and that's gonna imply having an exit and having as high a valuation as possible.
• 24:36
  SPEAKER_1: I
• 24:37
  SPEAKER_3: Yeah.
• 24:37
  SPEAKER_3: Yeah.
• 24:39
  I would without having thought about it that much,
• 24:41
  argue that ethical considerations go to the wayside.
• 24:45
  SPEAKER_1: You have a fiduciary duty to maximize return for shareholders, and the consequences be damned.
• 24:51
  SPEAKER_1: The second
• 24:53
  SPEAKER_3: Yeah.
• 24:53
  SPEAKER_1: model that we could imagine, for example, is a utility, right?
• 24:57
  SPEAKER_1: That it's owned publicly and you have users who are paying for it.
• 25:03
  And you can imagine another model similar to single-payer healthcare,
• 25:07
  where part of your wireless service is you pay a small fee for AI medical care.
• 25:15
  So each of those implies a different set or involves a different set of incentives
• 25:20
  that are centered around different moral choices, ethical perspectives.
• 25:26
  SPEAKER_1: I'm wondering if you have thoughts on that.
• 25:27
  SPEAKER_1: You know, you mentioned
• 25:28
  SPEAKER_3: Yeah, yeah.
• 25:29
  the health minister and 50 cents per citizen,
• 25:32
  for example,
• 25:34
  and you know,
• 25:34
  that funds a much different approach to the use of AI,
• 25:38
  I would think, than VC looking to cash out.
• 25:41
  SPEAKER_3: Absolutely.
• 25:42
  And and aga again,
• 25:43
  the three of us can sit in committee and and and suggest to 173 countries
• 25:48
  that there are these three,
• 25:50
  you know,
• 25:50
  models or maybe five models,
• 25:51
  but maybe ten models, and a recommendation which one they ought to pick.
• 25:56
  SPEAKER_3: And there's a United Nations badge on our front door and nobody cares.
• 25:59
  So it has to be done organically from within the culture's sensibilities
• 26:03
  of how they want to run it, which they will make mistakes, but that's what it is.
• 26:07
  SPEAKER_3: The benefits of of AI as as we see them are distributed basically for free.
• 26:13
  SPEAKER_3: So it's completely democratized to everybody.
• 26:16
  The only expense that they have is this $46 or whatever it ends up being,
• 26:20
  which they're gonna do anyway, because they need a cell phone.
• 26:22
  And the only question now is for free,
• 26:25
  you get all this power, AI power, that's hyper personalized to you.
• 26:28
  SPEAKER_3: Do you want it or not?
• 26:29
  SPEAKER_3: If the answer is no, then no.
• 26:31
  In Latin America,
• 26:32
  there is a project,
• 26:34
  and 12 nations have signed up to build their their sovereign AIs in the the style
• 26:39
  that I described.
• 26:40
  SPEAKER_3: Do you know the budget for the 12 countries?
• 26:43
  SPEAKER_3: $550,000.
• 26:43
  SPEAKER_3: That's kind of like a dinner bill for a medium-sized conference.
• 26:48
  SPEAKER_3: It's pathetic compared to the kind of civilizational questions you are posing.
• 26:54
  SPEAKER_3: So so my kind of solution is this.
• 26:56
  SPEAKER_3: Get a fund together, go into these countries.
• 26:58
  They probably each need twenty five,
• 27:00
  thirty five million to to mount the kind of effort it takes to organize
• 27:04
  that real model.
• 27:06
  SPEAKER_3: And here it is.
• 27:07
  But that has to be,
• 27:08
  to your earliest complaint,
• 27:10
  that has to be a transaction into a a private entity, not a government entity.
• 27:16
  SPEAKER_3: If it's a government entity, we have issues.
• 27:17
  It has to be private and with some government influence and participation for sure
• 27:23
  in authoritarian countries.
• 27:25
  SPEAKER_3: But it has to be fundamentally private, and the capital that goes in will look for an exit.
• 27:31
  SPEAKER_3: So that's the deal.
• 27:33
  SPEAKER_3: And that's a deal I'm perfectly happy to make.
• 27:36
  The institutional investors that I've spoken to who are sitting
• 27:39
  on nine point something trillion dollars in Luxembourg alone.
• 27:43
  SPEAKER_3: And the money sitting on the sidelines is looking for these kinds of things.
• 27:47
  SPEAKER_3: Some of them, the alternatives, are looking for these kinds of things.
• 27:50
  SPEAKER_3: We'll you take some, we'll apply to all these countries.
• 27:53
  SPEAKER_3: Good things and bad things will happen in those countries.
• 27:56
  SPEAKER_3: Why?
• 27:57
  SPEAKER_3: Because some people will grab the AI in the country and pervert it.
• 28:00
  SPEAKER_3: Okay, got that.
• 28:02
  SPEAKER_3: and it'll be a private sector transaction.
• 28:04
  SPEAKER_3: That's where I've place my bet.
• 28:05
  SPEAKER_3: With public participation.
• 28:07
  But not at the governance of the entity level,
• 28:10
  but simply at the you know, sort of at the at the margin, at the edges.
• 28:13
  SPEAKER_3: Where is where it belongs.
• 28:14
  Well with that mention of government Within the United States to narrow the focus,
• 28:18
  or pick any country, any jurisdiction.
• 28:20
  SPEAKER_1: What do you think the appropriate role of government is with respect to at least two things?
• 28:26
  SPEAKER_1: Number one, guardrails.
• 28:27
  You know,
• 28:28
  we have export controls,
• 28:29
  we have uses of certain kinds of tools and technologies that are outlawed.
• 28:33
  SPEAKER_1: And second of all, to mitigate the effects.
• 28:35
  SPEAKER_1: I mean, agriculture wrought it was extremely disrupted.
• 28:39
  SPEAKER_1: It was very lowercase v violent and every major technological inflection point has proven that.
• 28:46
  Some of that is about restricting certain uses,
• 28:49
  but the other is after the horse is out of the barn, how do you mitigate the harm?
• 28:52
  SPEAKER_1: W what is your view of government's appropriate role?
• 28:54
  SPEAKER_3: Okay.
• 28:56
  SPEAKER_3: Good.
• 28:56
  SPEAKER_3: Yeah.
• 28:57
  SPEAKER_3: So let's get let's get, you know, real, here.
• 29:00
  What I worry about the most is a hierarchy of of things that start on that side of
• 29:04
  the bracket of extinction.
• 29:05
  SPEAKER_3: And what can government do about those?
• 29:07
  What can government do about rogue agents that deep fake their way
• 29:10
  into people hating each other and starting the nuclear war that you referred to?
• 29:13
  SPEAKER_3: Let's start with things that are really, really momentous, that people should not do.
• 29:17
  SPEAKER_3: Well what role should government have when AI that is super intelligent.
• 29:21
  In other words,
• 29:22
  it's making its own goals that are,
• 29:24
  closely related to what the goals of the human is.
• 29:27
  SPEAKER_3: Then they make up their own.
• 29:28
  SPEAKER_3: And then they go learn, they code themselves to go learn what they need to learn.
• 29:31
  And then most of them are good,
• 29:33
  some of them are rogue, and the rogue ones are the ones I care about.
• 29:35
  SPEAKER_3: Criminals love AI.
• 29:38
  SPEAKER_3: my God, what a gift.
• 29:39
  SPEAKER_3: What a gift.
• 29:40
  SPEAKER_3: So make a list of criminality.
• 29:42
  SPEAKER_3: Just a long list.
• 29:43
  And the role of government should be to pre-think that,
• 29:46
  force in some way,
• 29:47
  whatever technological or legal or regulatory or even market constraint,
• 29:54
  restraint can mitigate those, and put people in jail.
• 29:58
  SPEAKER_3: And that's that's a proper role of government I would totally support.
• 30:03
  SPEAKER_3: And now, the one you mentioned is the subtle one that everybody's really interested in.
• 30:08
  For example,
• 30:09
  when millions and millions of people are thrown out of work
• 30:12
  because they're redundant,
• 30:13
  because AI does a better job with less errors,
• 30:16
  and and and in superintelligence,
• 30:18
  you can't tell if it's a human or not, and you sort of prefer the AI.
• 30:22
  So the future is what do you do with these hundreds of millions of people who
• 30:26
  who really didn't catch on to the need to understand AI on a personal level?
• 30:33
  SPEAKER_3: And when the thing went exponential, they cannot catch up.
• 30:36
  SPEAKER_3: They're done for now.
• 30:37
  So what happens when you have a large unemployed base of people in any country,
• 30:42
  particularly young youth, young males?
• 30:44
  SPEAKER_3: Not good things.
• 30:45
  SPEAKER_3: That's a bomb.
• 30:46
  SPEAKER_3: That's an absolute bomb in the center of any culture.
• 30:48
  SPEAKER_3: So that has to be addressed and mitigated.
• 30:51
  SPEAKER_3: Unanswered question, Tom.
• 30:52
  SPEAKER_3: And that's the nature of these mega paradigm shifts.
• 30:56
  SPEAKER_3: You sort of at first don't even know the question.
• 30:58
  SPEAKER_3: Don't even know the question, because it's so new and violent and strange.
• 31:02
  SPEAKER_0: Yeah and see
• 31:02
  SPEAKER_3: Right.
• 31:02
  SPEAKER_0: I don't I don't I don't think it's that hard.
• 31:04
  SPEAKER_0: I mean, I think there are some old bottles into which this new wine can be poured.
• 31:09
  SPEAKER_0: The New Deal is one model for it.
• 31:10
  SPEAKER_0: I I grew up in the Rust Belt, right?
• 31:13
  There were all sorts of things of FDR's first term that would have been solutions to
• 31:19
  to massive unemployment and which were stymied,
• 31:21
  and we we don't need to reinvent those kinds of things.
• 31:23
  SPEAKER_0: We
• 31:24
  SPEAKER_3: Yeah.
• 31:25
  SPEAKER_0: don't need to reinvent the notion of a safety net.
• 31:27
  SPEAKER_0: We don't need to reinvent the notion of social insurance.
• 31:28
  SPEAKER_0: We don't need
• 31:29
  SPEAKER_3: Yeah.
• 31:29
  SPEAKER_3: And I agree with that.
• 31:30
  SPEAKER_0: to re we don't need to reinvent any of those things.
• 31:32
  SPEAKER_0: We need to do them.
• 31:33
  SPEAKER_3: Yeah, yeah.
• 31:34
  SPEAKER_3: Now there's there's a big difference.
• 31:36
  SPEAKER_3: There's a big difference.
• 31:36
  SPEAKER_3: If if there's no change in the cognitive infrastructure because you're you don't have AI.
• 31:41
  If you're thrown out of work as a worker and then the economy comes back,
• 31:45
  you can go back to work.
• 31:46
  This one is if you miss the AI thing,
• 31:49
  you may not actually have the the skills to be interesting.
• 31:52
  SPEAKER_3: So it's a much worse problem than simply creating an FDR safety net.
• 31:56
  Second one,
• 31:56
  we do have an example of a country that has taken this thing very seriously,
• 32:01
  and it's called China.
• 32:03
  And China said a long time ago,
• 32:05
  we have to lift our base out of subsistence, and they did it pre-AI.
• 32:10
  Now they're pouring trillions of dollars into AI,
• 32:14
  into frontier AI, that's actually being done on a private basis.
• 32:18
  DeepSeek just announced that they're like what 10x 5x more more powerful and 1/5
• 32:24
  the cost or something of American AI.
• 32:26
  SPEAKER_3: I mean it's a it's an astonishing announcement.
• 32:29
  SPEAKER_3: Astonishing announcement.
• 32:32
  SPEAKER_3: and so China is pouring money into it.
• 32:34
  Their regulatory schema around that industry seems heavy-handed on the outside,
• 32:40
  but it's actually almost deregulated.
• 32:43
  They're pouring tons and tons of money into quantum technologies,
• 32:47
  which is which is the next discontinuity.
• 32:51
  SPEAKER_3: And we're not.
• 32:52
  They're pouring tons of money into fission and fusion,
• 32:56
  without which there's no electrons and without electrons there's no tokens and
• 32:59
  without tokens there's no nothing.
• 33:02
  SPEAKER_3: And by the way, fusion is that my other project is the great unlock for for abundance.
• 33:08
  SPEAKER_0: Yeah.
• 33:08
  SPEAKER_3: Yeah.
• 33:09
  SPEAKER_0: Yeah, no, you know, when I hear abundance I don't get my gun out, as Goring said.
• 33:13
  SPEAKER_0: I get my bullshit meter out.
• 33:14
  SPEAKER_0: And yeah.
• 33:15
  SPEAKER_3: Well wait a Michael, and now you've been pressing me.
• 33:16
  SPEAKER_3: Let me press you right back.
• 33:17
  SPEAKER_3: Please tell
• 33:18
  SPEAKER_0: Yeah.
• 33:19
  me how when energy is available in a distributed basis
• 33:25
  through micromodular fusion using essentially converted ocean water
• 33:30
  and has cheap electrons to make the water under your ground sweet,
• 33:34
  so you can drink it and you can then put that water in the ground and make food
• 33:38
  and take and take a big chunk of the cost out of household,
• 33:43
  energy cost, how th how that triggers your bullshit meter.
• 33:46
  SPEAKER_3: How does that not register in your mind is something like a wonderful thing?
• 33:50
  SPEAKER_0: If it could happen, I I agree.
• 33:52
  SPEAKER_0: I just don't, you know, I I don't see the key word, the key
• 33:55
  SPEAKER_3: Okay, so you're putting your butt you you're basically
• 33:57
  SPEAKER_0: word in your sentence was distribution, right?
• 33:58
  SPEAKER_0: And
• 33:59
  SPEAKER_3: Yeah.
• 33:59
  SPEAKER_3: Yeah.
• 34:00
  SPEAKER_0: and what what we what we always have is a problem there, right?
• 34:04
  SPEAKER_0: Is the distribution
• 34:05
  SPEAKER_3: Yeah, no, but
• 34:07
  SPEAKER_0: of this suddenly more available energy.
• 34:10
  Who controls that distribution and how does it get distributed,
• 34:13
  both physically and economically?
• 34:15
  SPEAKER_3: Okay, let's make let's make two bets and see if they're good.
• 34:18
  SPEAKER_3: Bet number one, fusion will not be solved in the next in ten years.
• 34:21
  SPEAKER_3: I I I have I have reasons to argue and I'm happy to argue.
• 34:24
  SPEAKER_0: Yeah, you no, I think I know you think it it's likely, right.
• 34:26
  SPEAKER_3: Yeah, yeah.
• 34:27
  And I and I can give you exactly the,
• 34:28
  you know,
• 34:28
  sort of the a through line that makes it quite plausible and probably
• 34:32
  and probably very likely.
• 34:35
  SPEAKER_3: You know, the first of a kind will be out there.
• 34:37
  Two, it's more efficient to make micromodular than to make these gigantic megaplanes
• 34:41
  for many reasons, not including the technology keeps moving.
• 34:46
  And anyone who knows anything about calculus,
• 34:48
  you don't want to invest in things that are going to be obsolete when
• 34:52
  the technology moves quickly.
• 34:53
  SPEAKER_3: You want to do it incrementally.
• 34:55
  SPEAKER_3: That naturally lends itself to small.
• 34:58
  SPEAKER_3: Third, resilience.
• 34:59
  It's really easy to drop drones that are brilliant,
• 35:02
  million drones, on a very large data center or energy plant.
• 35:07
  If you micromodular,
• 35:08
  it's seven cargo containers and you put them in the backyard
• 35:11
  of you know some somewhere.
• 35:13
  SPEAKER_3: Which which takes care of the AI.
• 35:16
  SPEAKER_3: That's a couple, two, three cargo containers, takes care of the energy.
• 35:21
  Initially fission,
• 35:22
  which I'm not thrilled about,
• 35:24
  but which is necessary,
• 35:25
  by the way,
• 35:25
  and fusion,
• 35:26
  which I'm very thrilled about,
• 35:28
  and some other things,
• 35:29
  cooling and batteries and controls and what you've got yourself a distributed system
• 35:33
  which is resilient and is a 25-year battery or 30-year battery,
• 35:36
  you know,
• 35:37
  it's actually longer than that,
• 35:39
  because you just keep pouring in relatively available fuel.
• 35:44
  But eventually the the low the low yield radioactivity will exhaust the the magnets
• 35:49
  and the lasers and all the things that are in the fusion containment chamber.
• 35:52
  SPEAKER_3: You have to replace it.
• 35:53
  SPEAKER_3: 25 years, you want to replace it.
• 35:55
  SPEAKER_3: So no big deal.
• 35:56
  SPEAKER_3: Now that's a scenario that says you gotta make a bet, Michael.
• 36:00
  SPEAKER_3: Number one, fusion will not be achieved.
• 36:04
  SPEAKER_3: and number two, that people will hew for all kinds of who knows superstitious reasons to big.
• 36:10
  SPEAKER_3: Big, you're right, it does not lend itself to distribution.
• 36:13
  SPEAKER_3: Small does.
• 36:14
  SPEAKER_3: incidentally, if you do small, guess what you don't have to do?
• 36:17
  SPEAKER_3: A grid.
• 36:17
  SPEAKER_3: Which is very expensive.
• 36:18
  SPEAKER_3: All grids, all grids
• 36:19
  SPEAKER_0: Hm.
• 36:20
  SPEAKER_0: Yeah, 'cause everything's local.
• 36:21
  SPEAKER_0: Yeah.
• 36:22
  SPEAKER_0: Okay.
• 36:22
  SPEAKER_0: Yeah.
• 36:23
  SPEAKER_3: were were were built on the wrong electrical technology.
• 36:25
  SPEAKER_3: So they all have to be kind of upgraded and replaced.
• 36:28
  The trillions of dollars it would take to actually modernize any electrical grid,
• 36:31
  put it into fusion.
• 36:32
  SPEAKER_3: Solve the whole problem locally, more or less.
• 36:34
  SPEAKER_3: You know, so
• 36:35
  Well, I think I I want to interject that it seems to me that part
• 36:38
  of what this discussion is animated by is a vision of technology,
• 36:42
  as empowering,
• 36:43
  as in liberating, as in something that advances individual human existence.
• 36:51
  And you know, Mike,
• 36:53
  as a counterpoint to your concern,
• 36:55
  it certainly seems like digital communication devices are extremely widely
• 37:01
  distributed.
• 37:02
  Now part of that is because there was government infrastructure,
• 37:06
  namely the internet, but through commerce billions of people
• 37:09
  SPEAKER_3: Mm.
• 37:09
  SPEAKER_3: Mm.
• 37:11
  SPEAKER_3: Yeah, you don't need the internet.
• 37:12
  SPEAKER_3: You don't need the internet for distributed cellular, to be clear.
• 37:15
  SPEAKER_1: Fair enough.
• 37:15
  But the concern I have with the future vision of fusion is based
• 37:24
  on my General Magic experience,
• 37:26
  that all of those benefits are orthogonal to a new type of problem that'll usher in.
• 37:33
  So I I don't dispute that it would be great not to pay pay my electric bill or worry
• 37:38
  about power outages.
• 37:40
  But when I'm pondering listening to the two of you engage in the discussion
• 37:44
  about it is fair enough,
• 37:46
  but what are we not getting about the new world that that would
• 37:49
  SPEAKER_0: Yeah.
• 37:50
  SPEAKER_1: usher in?
• 37:50
  SPEAKER_1: I think that that that's the sticking point I'm getting to.
• 37:51
  SPEAKER_1: Now maybe that's just a rationalization for being a Luddite,
• 37:54
  SPEAKER_3: Yeah.
• 37:54
  SPEAKER_3: So remember No no no no
• 37:57
  SPEAKER_1: but I I don't think so.
• 37:59
  SPEAKER_3: No, that's good, Tom.
• 38:00
  So so let me do what I said people should do and right at the head of this,
• 38:03
  you know, head end of this thing.
• 38:05
  SPEAKER_3: To be intellectually honest about the upside, you've got to embrace the downside.
• 38:10
  SPEAKER_3: So why don't I flip roles?
• 38:11
  SPEAKER_3: I'll flip roles.
• 38:12
  I'll go completely dystopian on you,
• 38:14
  and then invite you to argue with me right back from the,
• 38:17
  you know, from the positive side.
• 38:18
  SPEAKER_3: You're willing to play that game?
• 38:19
  SPEAKER_0: Sure, absolutely.
• 38:19
  SPEAKER_1: Let's do it.
• 38:20
  SPEAKER_3: Yeah.
• 38:20
  SPEAKER_3: You you but but you can't say, Yes, you're absolutely right, it's whole horrible.
• 38:22
  SPEAKER_3: You have to actually You have to play the game.
• 38:24
  SPEAKER_3: You have to play the game.
• 38:25
  No, no, no,
• 38:26
  and I look a I you know,
• 38:27
  I just put a column up I I put a column up about what the wonderful world
• 38:31
  of medical i innovation driven by AI.
• 38:34
  SPEAKER_0: I mean, I'm I'm not you know, I'm I I'm not ...
• 38:37
  SPEAKER_3: I'd be happy with your I'd be happy with your book if you actually took both sides.
• 38:40
  SPEAKER_3: And I think it'd be
• 38:41
  SPEAKER_0: Hmm.
• 38:41
  SPEAKER_0: Okay.
• 38:42
  SPEAKER_1: I want to remind Mike.
• 38:43
  SPEAKER_1: Marc,
• 38:45
  SPEAKER_0: Uh oh.
• 38:46
  SPEAKER_1: I don't think you know this.
• 38:47
  SPEAKER_1: When when Mike departed General Magic, one of the parting gifts was a homemade spinner.
• 38:54
  SPEAKER_1: And it just said at the very top, "Dr.
• 38:56
  SPEAKER_1: No."
• 38:56
  SPEAKER_1: No
• 38:58
  SPEAKER_3: Okay.
• 38:59
  SPEAKER_1: matter where the pointer landed, one was "Inconceivable."
• 39:01
  SPEAKER_3: Yeah.
• 39:03
  SPEAKER_1: One was "Why the F would we ever do that?"
• 39:07
  SPEAKER_1: Another was "That's the stupidest question I've heard all week."
• 39:11
  SPEAKER_1: I think another one was, "Who are you kidding?"
• 39:13
  SPEAKER_1: Another was, "Get out of
• 39:14
  SPEAKER_3: I love it.
• 39:15
  SPEAKER_1: my office." So Mike. So for
• 39:15
  SPEAKER_0: I s I still have I still have it.
• 39:17
  SPEAKER_3: Can you make a second copy and send it to my house?
• 39:19
  SPEAKER_3: I love it too.
• 39:20
  for Mike to be favorable,
• 39:20
  you know, for him to be bullish on something is really saying something.
• 39:23
  SPEAKER_1: And with that I will I will shut up and listen to your
• 39:25
  SPEAKER_3: Yeah.
• 39:25
  SPEAKER_3: But but that's a challenge for the next segment of this episode.
• 39:27
  SPEAKER_3: Is that's a challenge?
• 39:28
  SPEAKER_3: So you so you gotta jump out of your
• 39:30
  SPEAKER_1: Right. So let's hear your dystopian
• 39:30
  SPEAKER_0: Right.
• 39:30
  SPEAKER_0: So so
• 39:32
  SPEAKER_1: vision for us to refute.
• 39:32
  SPEAKER_0: no no but so the the the the downside of the energy abundance piece that you see coming.
• 39:34
  SPEAKER_0: What let's just no no no remember we're switching we're switching roles now, right?
• 39:37
  SPEAKER_0: So
• 39:38
  SPEAKER_3: You wanna give me the downside?
• 39:39
  SPEAKER_3: I was gonna give you the downside, but all right, keep going.
• 39:41
  SPEAKER_3: Okay, all right.
• 39:44
  SPEAKER_0: What are the possible risks of that world?
• 39:47
  SPEAKER_0: And what are the things that might prevent it from happening?
• 39:50
  SPEAKER_3: Yes.
• 39:50
  SPEAKER_3: I don't like abundant energy because (I'm I'm now in role, you know, so.)
• 39:54
  SPEAKER_0: Yeah.
• 39:55
  I don't like abundant energy of that sort,
• 39:57
  you know, which is kind of infinite and not expensive, because
• 40:00
  SPEAKER_0: You already prejudiced come on Marc.
• 40:01
  SPEAKER_0: Yeah, yeah.
• 40:01
  SPEAKER_0: Yeah.
• 40:01
  SPEAKER_0: No fair.
• 40:02
  SPEAKER_0: You gotta you gotta do it fairly, okay.
• 40:03
  SPEAKER_3: They teach you that in debate.
• 40:04
  SPEAKER_3: You know, you just
• 40:06
  SPEAKER_1: I don't like this delicious chocolate cake that's the best thing I've ever tasted.
• 40:07
  SPEAKER_1: Let me tell you what I don't like about it.
• 40:08
  SPEAKER_1: Okay, go ahead.
• 40:09
  SPEAKER_3: Because they teach you that.
• 40:11
  SPEAKER_3: They teach you that in the rhetoric.
• 40:14
  SPEAKER_3: Mike, Mike, Mike, it's rhetoric.
• 40:16
  SPEAKER_3: It's fine.
• 40:17
  SPEAKER_3: ...is that bad things will be cheaper to build.
• 40:20
  SPEAKER_3: And and that's that's the point.
• 40:22
  SPEAKER_3: So there will be you know, I'm making this up.
• 40:25
  There will be weapons that depend on energy and depend on all kinds of things to
• 40:34
  to shoot pulses of death rays at you.
• 40:37
  SPEAKER_3: And with with with really cheap energy, that's that's actually very viable.
• 40:41
  SPEAKER_3: Okay.
• 40:42
  You just carry along some of these cargo containers
• 40:44
  and you got yourself your own sort of amazing power plant,
• 40:47
  and you put a laser on it and you can shoot anything
• 40:49
  and kill anything like really effectively.
• 40:51
  SPEAKER_3: So anything that needs energy for bad things will do better.
• 40:56
  SPEAKER_3: It'll cost less.
• 40:57
  SPEAKER_3: It'll actually open up the business model, of that bad thing.
• 41:00
  Another bad thing,
• 41:01
  which is kind of an interesting you know, sort of ace of spades trump card here.
• 41:06
  SPEAKER_3: (I promised not to use that word.
• 41:09
  SPEAKER_3: It just makes me, you know, not not well.)
• 41:11
  SPEAKER_3: is is population explosion.
• 41:13
  SPEAKER_3: The carrying capacity of the world is already there.
• 41:16
  SPEAKER_3: And if you have abundance, it's you know, terrible.
• 41:18
  SPEAKER_3: Because when something is free or cheap and it's a good thing, people overconsume it.
• 41:22
  SPEAKER_3: We will absolutely overconsume it.
• 41:24
  SPEAKER_3: Whoopee, let's have 11 children.
• 41:26
  Because there's a safety net,
• 41:27
  so we have you know universal basic income, we have this, we have that.
• 41:31
  SPEAKER_3: Let's have 11 kids.
• 41:32
  SPEAKER_3: That will destroy the planet.
• 41:33
  But these kinds of consequences,
• 41:36
  to be intellectually honest,
• 41:36
  Michael, you have to sit down and brainstorm the red team torpedoes.
• 41:42
  And the reason I started, Tom,
• 41:43
  with with extinction level bad things is because you're looking for the Class
• 41:48
  A torpedoes that if they hit, you're dead.
• 41:51
  Class B torpedoes...
• 41:53
  Did you see,
• 41:53
  the movie with Tom Hanks where he was driving a a battleship cruiser,
• 41:59
  protecting a convoy in World War II.
• 42:01
  SPEAKER_3: And he had no air cover.
• 42:03
  It was in that zone of the Atlantic,
• 42:05
  and the the U 2 wolves came at them and were sinking ships right and left,
• 42:09
  and they came after him.
• 42:12
  SPEAKER_3: And and the sailor said, Torpedo!
• 42:14
  SPEAKER_3: And he went hard left, port rudder, starboard.
• 42:17
  SPEAKER_3: And and the and he was able to turn the ship just fast enough that it glanced off the side.
• 42:22
  SPEAKER_3: Class B torpedoes.
• 42:23
  SPEAKER_3: You can see them, they can be fatal, you can maneuver around them.
• 42:27
  SPEAKER_3: Class B.
• 42:28
  SPEAKER_3: Class C torpedoes is just chatter.
• 42:30
  SPEAKER_3: It's just people coming up with edge conditions.
• 42:32
  SPEAKER_3: And that's the majority of what I hear.
• 42:33
  SPEAKER_3: Chatter.
• 42:34
  SPEAKER_0: Mm.
• 42:35
  SPEAKER_3: It's it's very cheap and lazy.
• 42:37
  SPEAKER_3: It's like swearing.
• 42:38
  SPEAKER_3: If you don't want to articulate an emotion, you say, fuck.
• 42:40
  SPEAKER_3: No, you say I'm frustrated because of you know but but the lazy
• 42:43
  SPEAKER_1: Marc Marc.
• 42:44
  SPEAKER_3: the lazy approach is is is just to swear.
• 42:47
  SPEAKER_3: That's what I call Class C torpedoes.
• 42:49
  SPEAKER_3: Not not that interesting.
• 42:50
  SPEAKER_3: You just have to suppress
• 42:51
  SPEAKER_1: Mm.
• 42:52
  SPEAKER_3: Yeah.
• 42:52
  SPEAKER_3: Go ahead, Tom.
• 42:52
  SPEAKER_3: By the way, I didn't accuse you of class C torpedo.
• 42:54
  SPEAKER_1: I just never as as Mike knows, I never see a fucking problem with swearing.
• 42:58
  SPEAKER_1: But
• 42:59
  SPEAKER_3: huh.
• 42:59
  SPEAKER_1: no, that's that is I think a very prescient and and insightful way of prioritizing the risks.
• 43:07
  One one of the things I wanted to extract from what you were saying is
• 43:09
  the two crucial questions with respect to innovation:
• 43:13
  Number one, what does it facilitate?
• 43:15
  SPEAKER_1: What new kinds of things does it enable and to think as expansively about
• 43:19
  SPEAKER_3: Yeah.
• 43:19
  SPEAKER_3: Yeah.
• 43:20
  SPEAKER_1: that?
• 43:20
  SPEAKER_1: What you know, what constraints does it remove and think really broadly about that?
• 43:24
  SPEAKER_1: And second of all, what
• 43:25
  SPEAKER_3: Okay.
• 43:25
  SPEAKER_3: Yeah.
• 43:26
  kind of behavior does it incentivize,
• 43:29
  not just make possible, but make tempting in a way that was unthinkable before?
• 43:35
  SPEAKER_3: Okay, good.
• 43:36
  SPEAKER_3: Good.
• 43:36
  I'll make one last really dystopian point and then we'll we'll go
• 43:39
  into this really positive thing that you just asked about.
• 43:42
  The really dystopian thing is when intelligence,
• 43:46
  superintelligence, is cheap, it too will be overconsumed.
• 43:49
  If I'm a criminal and I can get my hands on a long list of really horrible things
• 43:55
  for cheap or free, I'm gonna overconsume that.
• 43:58
  So the dystopian argument is is there will be more criminals
• 44:01
  who do more criminal things in a smarter way.
• 44:04
  SPEAKER_0: Yeah.
• 44:05
  And and I thought,
• 44:06
  the whole notion of having a Me and having that Me wrangle all these agents
• 44:10
  for hire,
• 44:11
  including life coaches and companions,
• 44:14
  I see that as the destructiveness of social media on steroids
• 44:17
  that it will fundamentally alienate kids as they grow.
• 44:19
  You know,
• 44:19
  you talked about kids being super intelligent natives the way
• 44:24
  that they're becoming they're digital natives now.
• 44:26
  I can't imagine a world in which children grow up having their phone be
• 44:30
  the most important thing in their life,
• 44:32
  not just because it lets them hook up with their friends,
• 44:35
  but because it it it gives them all the emotional sustenance they need.
• 44:41
  SPEAKER_3: Let let me let me go into that future.
• 44:45
  SPEAKER_3: This is the not yet known, the kind of glimmers, you know, What good things could be out there?
• 44:51
  SPEAKER_3: When I'm out there, what I see first and foremost is science.
• 44:55
  SPEAKER_3: as the great unlock.
• 44:57
  SPEAKER_3: And what I see as pernicious or potentially pernicious is cognitive, the large language model.
• 45:02
  SPEAKER_3: Why?
• 45:03
  Because when you have a mega, mega,
• 45:06
  mega amplifier of the sort we've never seen before,
• 45:09
  it mega amplifies every part of the human condition.
• 45:12
  SPEAKER_3: So to the extent people are horrible, they'll be mega horrible.
• 45:16
  SPEAKER_3: To the extent people are wonderful, they'll be mega wonderful.
• 45:19
  And and this is the greatest amplifier,
• 45:20
  cognitive amplifier ever, for those who participate in it.
• 45:24
  SPEAKER_3: Not everyone will participate in it.
• 45:26
  So And and there will be all kinds of inventions and and brilliant things
• 45:31
  that happen when people's minds are magnified.
• 45:34
  You know,
• 45:34
  and that could be a zero-sum game where there'll be
• 45:36
  as many horrible people magnifying as wonderful.
• 45:39
  SPEAKER_3: I'm on the side there'll be a lot of wonderful things done.
• 45:43
  SPEAKER_3: You know, my my hero Demis, just for a moment.
• 45:45
  There was a meeting in January,
• 45:49
  and there were people in the Vatican,
• 45:50
  you know,
• 45:50
  from the Vatican sitting in the room,
• 45:52
  room of maybe 80 people,
• 45:55
  and and everyone else from from the people that you would view as the elite of
• 45:59
  the elite of AI that are the enemy of the people.
• 46:01
  SPEAKER_3: They were all there.
• 46:04
  And Demis said,
• 46:05
  you know,
• 46:07
  what what we're doing is we're now digging into the very foundations of everything
• 46:12
  that the universe is built out of.
• 46:14
  SPEAKER_3: Atoms and subatoms, molecules of various sorts that do not exist yet in nature.
• 46:18
  SPEAKER_3: And we are trying to do things with them that are that are beneficial.
• 46:23
  SPEAKER_3: Beneficial.
• 46:23
  SPEAKER_3: Medical is one class of examples, materials is another class of examples.
• 46:28
  SPEAKER_3: That'll happen.
• 46:28
  SPEAKER_3: Science will be revolutionized.
• 46:30
  And what he said in the presence of these Vatican people is:
• 46:34
  What we all are reaching at is is the mind of God.
• 46:39
  SPEAKER_3: You know, jaws dropped.
• 46:41
  SPEAKER_3: It's like, what?
• 46:42
  SPEAKER_3: You know, what is he saying?
• 46:44
  Well, what he's saying is that the Vatican people are trying to understand
• 46:48
  the deepest mysteries of the universe.
• 46:51
  SPEAKER_3: through their, profound belief in in how that happened.
• 46:55
  And the philosophers and humanists are trying to understand it,
• 46:58
  in ideological terms, human nature terms, social organization terms.
• 47:02
  SPEAKER_3: And we scientists are trying to say, okay, subparticles.
• 47:05
  SPEAKER_3: what's what's what's inside the atoms and when you crack them?
• 47:10
  And then how do we combine them in interesting ways to do things that God hasn't
• 47:13
  yet figured out?
• 47:14
  SPEAKER_3: Because God uses nature as his as you know, as his lab.
• 47:17
  SPEAKER_3: And nature takes its time.
• 47:19
  SPEAKER_3: But we we don't want to we're we're impatient.
• 47:21
  So we want to do in in quantum calculations and other kinds of calculations,
• 47:25
  you know,
• 47:25
  they're only done with AI,
• 47:27
  on chips that don't yet exist,
• 47:28
  by the way, to come up with things that are like amazing.
• 47:31
  SPEAKER_3: Amazing.
• 47:31
  So, you know,
• 47:32
  the punchline to every joke
• 47:34
  about people like us is you're gonna cure you know world hunger cancer and war?
• 47:39
  SPEAKER_3: The answer is yeah.
• 47:40
  SPEAKER_3: Yeah.
• 47:40
  SPEAKER_3: It's exactly where we're headed.
• 47:42
  SPEAKER_3: And I and I I I don't belong in that class because they're actually doing it.
• 47:45
  SPEAKER_3: I'm just talking about it.
• 47:47
  SPEAKER_3: But but that's where I live.
• 47:49
  I believe in God,
• 47:50
  but my God is a science,
• 47:51
  and my science is rooted to using advanced AI that we're just beginning
• 47:55
  to understand now.
• 47:56
  SPEAKER_3: I'm very optimistic.
• 47:58
  SPEAKER_3: I have one I have one more comment I want to make.
• 48:00
  SPEAKER_3: The comment I want to make is the peak experience of my life was shared with the two of you.
• 48:06
  SPEAKER_3: And shared with the two of you and others.
• 48:08
  And I just wanted to pause to thank the people that came on that journey
• 48:12
  because they they had enough of their own vision that when a vision was presented
• 48:16
  to them, they said, that's my vision too.
• 48:18
  SPEAKER_3: And that's what we all did.
• 48:19
  SPEAKER_3: So Mike, and Hersh, you know, thank you for being on that journey.
• 48:23
  And here we are,
• 48:23
  decades later,
• 48:24
  talking about yet another revolution,
• 48:26
  which is a a million times more powerful than the one we did.
• 48:30
  However,
• 48:30
  as we said on this on this show,
• 48:32
  the democratization of that incredible revolution is right back to what we did.
• 48:36
  SPEAKER_3: Smartphone.
• 48:38
  SPEAKER_3: How how about that for a full circle irony?
• 48:40
  SPEAKER_0: That's good.
• 48:41
  SPEAKER_0: Yeah.
• 48:42
  SPEAKER_0: And it wa it was it was the most life changing experience I ever had.
• 48:45
  SPEAKER_0: and
• 48:46
  SPEAKER_3: Me too.
• 48:46
  SPEAKER_3: Me too.
• 48:47
  SPEAKER_0: it stayed with us.
• 48:47
  SPEAKER_0: Here we are thirty years later.
• 48:48
  SPEAKER_0: And
• 48:49
  SPEAKER_3: And here we are.
• 48:49
  SPEAKER_3: Here we are.
• 48:50
  SPEAKER_3: Talking about the next revolution.
• 48:52
  SPEAKER_0: Yeah, and how to get it right this time.
• 48:53
  How to get it right this time of in the beginning of this revolution
• 48:57
  when we are not entitled to actually know what the questions are.
• 49:01
  SPEAKER_0: Well again, I want to thank thank you, Marc, for taking the time to do this.
• 49:03
  SPEAKER_1: Yeah, Marc.
• 49:03
  SPEAKER_1: It's always it always is mind expansive.
• 49:07
  SPEAKER_0: It's always a pleasure to hear you talk about the things that might be.
• 49:10
  SPEAKER_0: You know, it got me thirty years ago.
• 49:13
  SPEAKER_0: It it gets me now, you know.
• 49:14
  SPEAKER_0: So thank you.
• 49:15
  SPEAKER_3: Lovely.
• 49:15
  Well, thank you for the opportunity and Tammie,
• 49:17
  thank you for for for for it also and everybody else behind the camera there
• 49:20
  that I don't see.
• 49:21
  SPEAKER_3: Thanks everyone.
• 49:22
  SPEAKER_1: Excellent.
• 49:22
  SPEAKER_1: Thanks, Marc.
• 49:23
  SPEAKER_2: Thank you for listening to General Magic and the Soul of Tech podcast.
• 49:27
  Tune into our Substack for more interviews and articles
• 49:30
  at spellboundproductions.substack .
• 49:32
  SPEAKER_2: com.