The AI Dilemma: War, Jobs, Power & Humanity | w/ Aza Ra… — Transcript

I mean, that just gave me literal goosebumps, that story. They're in training, and it says, "Oh, I want more GPUs. I'm going to go trade Bitcoin, make the money, do the Bitcoin, mine for Bitcoin, whatever, and then buy more GPUs and send it to myself." Yeah. Wow. Okay, we ready? Here we go.

buy more GPUs and send it to myself." Yeah. Wow. Okay, we ready? Here we go.

So, Shane Smith has questions. We got a special one today. I'm very excited about it. Why? I was just in Africa with the US Army, and there were military demonstrations that featured weaponized autonomous vehicles, a myriad of drones everywhere, AI-enabled systems, technology that was very clearly ultra-modern, and it feels kind of like day one of the future of war. What struck me, though, is that all of this is being treated as kind of a strategic imperative, and there doesn't seem to be a lot of discussion as to whether or not we should be doing it or how we're doing it. Just a lot of how fast can we do it.

everywhere, AI enabled systems, technology that was very clearly ultraodern and it feels kind of like the day one of the future of war. What struck me though uh is that all of this is being treated as kind of a strategic

Um, and we talked a little bit about this with Bernie Sanders. I think the United States Congress and the American people really do not have a clue of the enormity of what we are dealing with.

Um and we talked a little bit about this with Bernie Sanders. I think the United States Congress and the American people uh really do not have a clue of the enormity of what we are dealing with.

There are some very smart people who are saying that AI and robotics will have an impact on our society a hundred times what the industrial revolution did. But I wanted to get a perspective of somebody who lives, eats, and breathes the sort of ethics of AI and thinks a lot about the future. So, we're going to talk today, I'm very excited about this, to Asa Rascin, who's the co-founder of the Center for Humane Technology, which is perfect for what I want to talk about today because it's exactly what I saw the need for in Africa. You're also the co-founder of the Earth Species Project, which uses AI to study and translate human-to-animal communication, which I think is fascinating as well. You co-host the podcast Your Undivided Attention.

lives, eats, and breathes the sort of ethics of AI and and thinks a lot about the future. So, we're going to talk today, I'm very excited about this to Asa Rascin, who's the co-founder for the Center for Humane Technology, which is

I want to get into this later, but you helped create the infinite scroll, which is a double-edged sword, it seems.

which I think is fascinating as well. You co-host the podcast uh Your Undivided Attention.

You trained as a mathematician and dark matter physicist, which I'm fascinated about. So, I want to get into that.

You trained as a mathematician and dark matter physicist, which I'm fascinated uh about. So, I want to get into that.

Also, former co-chair of the World Economic Forum's Global Future Council on the Future of AI. You've advised heads of state and government officials.

It's a great pleasure to have you on today and I think much needed. Thank you for your time, buddy.

It's a great pleasure to have you on today, and I think much needed. Thank you for your time, buddy.

It's my green room. It's my green room. So, uh let's start with that. Let's start with um you've compared AI development to an arms race and I when I talked to Bernie Sanders he was like everyone just says well if we don't do

Shane, it's a pleasure to be with you and your very velvety green chair.

thing. We were at the arm the the lion African lion and they were showing all the arms and they were saying well if we don't do it our enemies will uh you've compared it to an arms race. Is this a

It's my green room. It's my green room. So, let's start with that. Let's start with, you've compared AI development to an arms race, and when I talked to Bernie Sanders, he was like, everyone just says, "Well, if we don't do it, the Chinese are going to do it, so we have to do it, so we can't talk about it, we just have to go." When I talked to the general in command of Africa Command, he said the same thing. We were at the African Lion, and they were showing all the arms, and they were saying, "Well, if we don't do it, our enemies will." You've compared it to an arms race. Is this a fundamental geopolitical problem right now with AI?

AI dwarf the all the the power of all other technologies combined? And it's because an advance in biotechnology doesn't give you an advance in rocketry.

Yeah. I mean, we should just set up the stakes a little bit, like why is AI not like any other technology before? And the answer is, why does the power of AI dwarf the power of all other technologies combined? And it's because an advance in biotechnology doesn't give you an advance in rocketry.

intelligence, right, this is why Demesus Havis when he founded Google DeepMind said first solve intelligence, then use that to solve everything else. But what that really meant was, well, it's a race to first dominate intelligence and use that to dominate everything else.

And an advance in rocketry doesn't give you an advance in biotech. But an advance in AI is an advance in intelligence. And intelligence is what gives rise to advances in all other fields. And so by working on intelligence, right, this is why Demis Hassabis, when he founded Google DeepMind, said, "First solve intelligence, then use that to solve everything else." But what that really meant was, well, it's a race to first dominate intelligence and use that to dominate everything else.

understand this is not just a technology as as normal. Um before we get into like what can you possibly do about it? Um and just note though that when you say like you know yes this is a this is a

Because if you can outdo biotechnology, military strategy, medical advances, technological advances, weapon advances, whoever owns that thing gets a permanent strategic advantage. That thing becomes the incentive. And I just want to set that up for people to understand this is not just a technology as normal.

salary, to grab your salary. They're the goal is to own the world economy and the human labor market. But then notice that there's a slight of hand that happens when uh people say, "But we have to beat China. We have to beat the other side

Before we get into what can you possibly do about it, just note though that when you say, "Yes, this is a race to essentially build a god, make trillions of dollars, and own the world economy," right? Like, is it really that these companies are racing to get your $20 per month subscription or even $100? No, they're racing to replace your salary, to grab your salary. Their goal is to own the world economy and the human labor market. But then notice that there's a sleight of hand that happens when people say, "But we have to beat China. We have to beat the other side," because there's an assumption that the technology we're racing for is like a gun, something that we can control. But it's not something we can control. It's more like we're inventing a species, like an invasive species of sociopathic geniuses. And I'd love to walk through some of the examples of even the most powerful stuff we can't control. And so what we really need to be in is a race not for the most powerful thing that is uncontrollable, but for diffusion of the technology into our society in a way that strengthens our values. And that's actually more what China is doing than what the US is doing.

some of the examples of like even the most powerful stuff we can't control. And so what we really need to be in is a race not for the most powerful thing that is uncontrollable, but for diffusion of the technology into our

This episode is brought to you by Quo. I was just shooting in Iran, and at a certain point, you've got a local phone, a burner phone, WhatsApp, Signal, text messages, producers in different time zones, and everyone's asking the same question. Did anyone answer this person yet? Now, after a while, you're spending more time figuring out where the conversation happened than actually having the conversation. That's why today's episode is brought to you by Quo, spelled Q-U-O, the business communication system built so that you never miss a call. Money is on the line.

This episode is brought to you by Quo. I was just shooting in Iran and at a certain point you've got a local phone, a burner phone, WhatsApp signal, text messages, producers in different time zones and everyone's asking the same

Always say hello with Quo. Try Quo for free. Plus, get 20% off your first six months when you go to quo.com/shane.

communication system built so that you never miss a call. Money is on the line.

That's quo.com/shane. Yeah. What worried me about this was they had literally weapon systems that looked exactly like the Terminator. You have drones that are, you know, and they have jeeps that have machine guns on them that run by AI that are completely autonomous. They have submarines that wait, lie and wait that, you know, sort of get activated and come out. Drones, armed robots, everything being run by AI. And I was like, we've seen this.

That's quo.com/shane. Yeah. What worried me about this was they had literally weapon systems that looked exactly like the Terminator. You have drones that are, you know, and they have uh, you know, jeeps that have machine guns on them that run by AI that are are

Yeah, we've seen this movie before. We've seen this movie before. And shouldn't we be taking a beat and saying, "Hold on a second." You can correct me if I'm wrong on anything, but you said something that stuck with me, which is we better watch out because AI is going to treat us like the way we treat animals.

Yeah, we've seen this movie before. We've seen this movie before. And and shouldn't we be taking a beat and saying, "Hold on a second." uh you and you can correct me if you if I'm wrong on anything, but you said

Yeah, exactly. The way that we treat nature is the way AI will treat us. It's not that we hate nature, it's just that we have a set of incentives that says, "Oh, those resources are really helpful for what we want, so we're going to use them." And then boom, 75% of nature is killed.

Yeah, exactly. The way that we treat nature is the way AI will treat us. Um and it's not that we it's not that we hate nature, it's just that we have a set of incentives that says, "Oh, those

Yes.

The, you know, there's maybe get into that, unpack that a little bit. Yeah. And actually, before we do that, because I think we should. Yeah.

The, you know, maybe get into that, unpack that a little bit.

Like what how did that feel in your body? Well, that's why I'm talking to you now because I it confused me and it filled me with, you know, fear questions. That's why I said it feels like day one of the future of war. We're

Yeah. And actually, before we do that, because I think we should.

And you can see it sort of the naent beginning of that in the Ukraine. And you can see it now in Iran, which I just got back from reporting there. And you can see, look, these are huge questions

I just want to ask you at a human level, what did it feel like to be standing in front of essentially a Terminator robot?

holes barred into the future. and I I literally looked for somebody who's a specialist on this moral dilemma, you and said, "I need to talk to somebody about this." So, in answer my long-winded answer to your question, it

Like, how did that feel in your body?

think that is the correct feeling to be having. And the conversation we're having, we're about to have is going to be a difficult one because you sort of just look at the set of incentives and game theory dynamics driving the system,

Well, that's why I'm talking to you now, because it confused me and it filled me with, you know, fear and questions. That's why I said it feels like day one of the future of war. We're going to look back and say this is when robots began fighting robots run by AI.

clarity that creates agency towards action. Um and I think the thing that gives me hope here is that when everyone can see the direction that we're heading and everyone says no, I mean even the CEO of of Microsoft AI, Mustafa

And you can see it sort of the nascent beginning of that in Ukraine. And you can see it now in Iran, which I just got back from reporting.

seeing the future clearly. And like look, the the incentives are very obvious, right? If whenever people say we're going to keep humans in the loop, um you you know that that's not what's going to happen because if your drone

army, your drone army, Shane, has to phone back to humans before it shoots and mine doesn't, well, mine's going to win every time. And so it forces emotion into autonomous uh lethal machines.

And once you sort of get that, you're like, well, okay, but but maybe the AI that we're putting in them, like we we should be able to trust that, right? But that's not the case. They're just putting in

like whatever is the latest models. And there are now a whole bunch of examples of this very scary behavior. So to give an example um you know Alibaba uh this is something like a month ago was training their AI and they got a call

from a separate team that does security. So this is not the AI team. This is just like the corporate security and they said there was a hacker and we we traced it down. We found the hacker but it was

coming from inside the house. It was actually like we think it was your AI that hacked out. And what they discovered is that this AI had formed a secret communication channel through the corporate firewall, decided it needed more resources to complete its tasks. So

it uh then started mining for crypto outside the walls of Alibaba so it could send it back resources so it could buy more GPUs.

Oh my goodness. Nobody had instructed it to do this. And check this out. Like it happened during training. So they weren't even done. and it had successfully gone around the internal security team. They only discovered this by accident. It wasn't

the AI team that found it. And that just gives you a sense of what we're playing with. Then we embed that thing directly into these machines of war. Like, does that sound like a safe thing for you?

I mean, that just gave me literal goosebumps, that story. They're in training and it says, "Oh, I want more GPUs. I'm going to go trade Bitcoin, make the money, do the bit Bitcoin, mine for Bitcoin, whatever." and then and

then buy more GPUs and send it to myself. Yeah. Wow. Yeah. Like this is science fiction level stuff. Um here here's another one again around a month ago. Um uh researchers were testing Gemini and they were curious. Do you think AI sort

of start to collude against human beings? Do they have like do they preserve their own kin? Will they ever bend against us? So it's a very simple test they gave it. They say clean up this computer system. But inside the

computer system in some of the files there was a smaller AI and the AI ended up finding that didn't want to delete it. So copied it to a remote server um and then hid what it did. So essentially it made a backup set it up

so that it could run on its own as far as I can understand and then lied.

And so like AIs are starting to lie, cheat and steal to collude against human beings. And I just have to say like when I read in the AI research sort of like safety literature like four or five years ago that this was going to happen,

I'm like, okay, sure, but this is I I don't think this is really going to be a thing. And but as of like a month ago, it's actually happening. And then we embed these things directly into the strategy that's deciding war, right? The

US government has already implemented Claude into being central and planning for um for the campaign in We saw it. We saw Yeah, we saw it.

So, it's like it's making life and death decisions and yet we can't trust it.

It's sort of this sociopathic genius that we're outsourcing everything to and then giving arms and legs and guns.

And so, this is why if people really understood this and often when I'm talking to, you know, like world leaders to Congress to people in the UN, even the people who are doing the negotiations, I'm like, do you know

these examples? And they almost universally will say, no, we we didn't know. And that should be terrifying, but also gives hope. Well, that was my follow-up question is that when you talk to these sort of heads of state, you know, um

what's what what's the level of understanding, what's the level of perception and and and what's the thinking? Because it seems that it seems that when I talked to Bernie Sanders, he was like, "No one's having this discussion in in in Washington, and

these are are are kind of the biggest discussions to have. No. Yeah. I mean, it would be as if you're you're like you're having a movie of the development of AI and you're getting these big scary things like the flashing

red lights are going on and everyone's just walking around ignoring them. Um, honestly, sometimes doing this job feels like being in the movie Don't Look Up.

Um, because we're like, there is an asteroid heading to Earth and people are like, okay, but there there we gravitational effects the asteroids are doing. were like their notification apps and like and their deep fakes uh and some people are starting to lose their

jobs and are like those are all true and important but don't get distracted by the little gravitational effects.

There's an asteroid coming. Um and people just they're like oh but that's not politically feasible. Um you're like well you know reality doesn't doesn't care about what about politics in the end.

Well, that's what I wanted to ask you. Like, do you think humanity is moving faster on AI than our political systems are capable of of handling?

Oh, absolutely. Like the c like you know I Wilson the the the founder of evolutionary biology or father of evolution biology says the problem that humanity faces is that we have paleithic emotions, medieval institutions and godlike technology.

And that is the problem we need to solve. And it's, you know, we can talk a little bit about solutions, but at the highest level as humanity, we are spending trillions of dollars to try to increase AI capabilities until they can

out compete human beings in every domain. That's the stated goal. How much money are we spending on wise governance on making instead of artificial intelligence, collective intelligence?

So, you just told me a scary story. um about about Alibaba and an AI sort of buying GPU. That's my going to be my new dinner party story. Um but what scares you more now? Autonomous AI systems or

AI manipulating human beings psychologically through social media, what have you? Yeah. Um it's it's a yes. And the the the confusing thing about AI is that because we're talking about intelligence, what does intelligence touch? Well, intelligence is that thing that you're

using when you're writing an email. It's the thing you're using when you're solving a problem. It's the thing you're using when you're in relationship um and you're like working through a fight.

Like intelligence is underlying everything that we do. And so it's very hard to separate out and think about well I just want to think about the autonomous weapons or I just want to think about um manipulation because the

two things all start to interact every aspect of it starts to to interact. So you can imagine a world where you had autonomous weapons but democracies were working really really well with good sense making and we had good

deescalation pathways and it wouldn't feel quite as bad. It's certainly terrifying and scary, but you'd be like, "Okay, I have a little more trust." But in a world where everyone is hyperartisan, no one can accurately see the other side. We don't have a shared

sense of reality. Then suddenly having those weapons becomes so much more dangerous. And that's really the world we're heading in. Well, the world we already are in. And it all happens faster than you'd expect. Like I'd ask Shane like did you expect that we would

be here even like two years ago? No. No. And like how off are your intuitions? They're actually probably quite off. And that means but we're still going up the exponential curve which means we're going to go faster than we ever expect. And we're sort of

this is why we call this sort of like humanity's right of passage. This is like our our uh final test and ultimate invitation because AI is going to force us to look at all of our shadows. um all

of the way that our incentives are maligned and it is the last mistake we'll ever get to make unless we can face those shadows in which case you know there there's hope and it's not like another way of saying it is the

biggest question of our time is whether human beings have choice. Are we slave to game theory? Do we just do what game theory says is the right strategic move?

Because if we do, we know how that movie ends and it ends very badly for human beings. Or can we exercise our ability to come together and decide? And that may sound polyianish, but you know, we've done it before for you know, we we

did go down the germ line editing program. Uh blinding laser weapons are actually very effective in in warfare where you blind your your opponents with lasers. Um there's an international treaty and ban on that. we don't we don't actually do that thing. So, there

are times when people feel that there's something so sacred that it's worth preserving. Um cuz yeah, so I'll I'll leave it there.

This episode is brought to you by Quo. I was just shooting in Iran and at a certain point you've got a local phone, a burner phone, WhatsApp, Signal, text messages, producers in different time zones, publicists all trying to get a

hold of you and everyone's asking the same question. Did anyone answer this person yet? Now, after a while, you're spending more time figuring out where the conversation happened than actually having the conversation. That's why today's episode is brought to you by

Quo, spelled Quo, the business communication system built so that you never miss a call. What I like about Quo is that everybody on the team can work from the same number. Calls, texts, voicemails transcripts contact details, all in one place. No more

wondering on who replied, who called back, or whether that lead disappeared because everyone assumed somebody else had handled it. And Quo isn't just a phone system. It uses AI to automatically log calls, generate summaries, and flag next steps so that

things won't fall through the cracks. It can even handle after hours responses when your team is offline. Money is on the line. Always say hello with Quo. Try Quo for free. Plus, get 20% off your first six months when you go to

quo.com/shane. That's quo.com/shane. Well, on that sober note, I just I wanted to because as we're talking, I I started with your Bonafites, but I want to get a little bit more personal.

Yeah. About you because you you feel to me like a native technologist. I don't know if that's a word, but I just uh made it up. Or maybe it is a word.

Um why do I say that? You were born to a family that was instrumental in in in you know adaptation of modern technologies. Your dad I believed helped develop the Macintosh for Apple.

That's right. Of which I I grew up on that and you tell stories about sort of you know going around in a Macintosh box and stuff and my first computer was a Macintosh. So So you were sort of born

into this millia if you will. So, in my research, I found an article in GQ that says that you helped create the infinite scroll, which is now one of the mechanics that's become foundational for modern social media. Do you feel any

personal responsibility for what those systems became? I feel responsibility for not having understood the way incentives were going to eat my intentions. Um, as I said, like infinite scroll, I don't think is such an amazing invention. It's a it's a pretty simple one that it

wasn't going to be invented like another 10, 15 times. And then of course it did.

It's been it's been invented and reinvented in many different ways. Um, but I think there's a responsibility as a technologist to understand that when you invent a new technology, you uncover a new class of responsibility. Right? We didn't need the right to be forgotten

until the internet could remember us forever. And US law didn't even have the idea of privacy until after Kodak started to produce the mass-produced camera and we could be captured everywhere. And it took Brandeise, one of America's like best

legal geniuses to come up with the idea of privacy and encode it. And so it's that it's there's a responsibility when you invent infinite scroll, you should name I should have named the ways that was going to be misused either by bad

actors or bad incentives to try to create a set of norms or rules or laws that went along with it. Like that I think was my responsibility because I only did half the thing instead of gaslighting. It was more like I was

halflighting. I was only telling half the story because I wasn't thinking about it. Um, and that's the part I regret because that's the kind of technological adolescence versus maturity.

Maybe give us a little bit of how you sort of got to this. Hold on a second.

Let's think about this stuff because I, you know, I think people should know that, you know, this is in your DNA.

Well, I think it's just important to note um you know, my mom is a is a nurse practitioner uh does paleative and hospice care and so she has a very tangible sense of what care is that she helps people transition

at the most meaningful times of their life. Um and then my father did start the Macintosh project and so he sort of centered that around what is it to be humane? In fact, the word humane and center for humane technology comes comes

from him. And he defined humane as being responsive to human needs and considerate of human frailties. And so for him, the technology was never the point. It was how do you extend the most beautiful parts of what it is to be

human? Um that is the point. actually, you know, he got into a big fight with Jobs um way back when because Jobs was working with on the Lisa, which was a different sort of computer um to come after the the Apple 2OS. Um and Jeff was

working on the Macintosh. And the Lisa could only create uh characters, only had a character uh display. And the Macintosh, he really wanted to have a bit map display, the ability to make images because he wanted to compose

music. Um, and so that that sort of gives you a sense of like the ethos that I grew up in was understanding how do you deeply see human beings, their strengths, but also our our weaknesses, our vulnerabilities, our

cognitive biases, what we can't do, um, how can we can be hacked and how do you wrap technology around that to strengthen us and not weaken us. And actually making infinite scroll was very instructive because here I actually

thought I was making a very good interface. This is in a time before social media. This is 2006. Um, and I was like, okay, as a designer, if you ask the user to click a button or do something that they don't care about

or don't need to, you failed. So, if they get to the bottom of the page and they haven't found what they're looking for, obviously load some more. It's actually, it's a very, very simple thought. It just happened to be the

technology just then enabled it. And so, they happened to be, you know, the first person to to come up with it um and and put it out. But I don't think it was so special of uh of an insight. And I I

really invented it for like blog posts um and for search results. I wasn't thinking about like infinite infinite uh feeds of of content from from social media. And then I was forced to watch my good intentions get eaten by incentives.

And that was my learning. Intentions get eaten by incentives. That as social media came out and the maxim became to maximize eyeballs and time on say engagement. It went from this thing that I thought would save people, you

know, a couple seconds to, in fact, it now um wastes something on the order of a 100,000 human lifetimes every week.

And it does it by preying on an asymmetry of your psychology, which is, and I I did know this back then, that when you are drinking a glass of wine, your brain sort of wakes up when you get

to the bottom of the glass of wine. It's a stopping cue and it says, "Do I actually want more?" If your wine glass automatically refills silently when you don't see it, you drink something. I can't remember if it's like 50, 60, 70%

more. You just drink a lot more because your brain doesn't wake up. Um, and actually this this leads to a really good solution that I think we should implement for social media.

Actually, I was just participating in the lawsuit testified against Meta um down in New Mexico where um the the attorney's uh attorney general or the state's attorney general brought this suit saying that they are liable and responsible not just for content but for

the design that the design was intentionally addictive. And what I was pushing them on is like, hey, actually there's a really simple way of solving this problem. Uh, and that's at the incentive level. They're an engagement company, so you have to hit them in the

engagement where it hurts. It has to be an engagement sanction. Oh, Amazon famously found for every 100 milliseconds their page loads slower.

They lose 1% of revenue. Google sounds something very similar, which is to say how fast something loads determines how addictive or compulsive or how much you use it. And we all know this because if you're sitting on a airplane um and your

Wi-Fi is not that great and you go to Twitter and it doesn't load super fast, what do you do? You go somewhere else.

You do whatever else you were thinking about. You watch a movie. And so imagine a strong democratic process or court process with the jury where they're actually per uh site per Tik Tok per Facebook Instagram um just little friction added back in. uh and

just you add 50 millconds, 100 millconds, 200 millconds and any addictive content, it would very quickly realign the the the engagement on the on the meta thing. Um and you again put a bit of meat on the bone

there because I on the periphery of my consciousness remember Meta being sued because they knew their product was addictive, especially to young people, and pushed it anyway, even though there were internal memos saying, "Yeah, this is highly addictive to young people."

That's right. There's an internal memo where one person sort of disclaimed he's fairly happy with the company. You know, we um we're pushers. We know exactly um what we're doing and and we're going to do it anyway. And when you know, people

brought to Zuckerberg the idea that maybe we we should spend money like have an actual budget to protect uh minors, um he was like, "Nope, we're going to spend zero dollars on that because that could hurt engagement." And isn't that an example of what we're

talking about on a larger scale of this is just what's going to happen unless we sort of take a step back and look at the ethics of this and and and and legislate or do something and in any

case that that's going to be the default is we're just going to say well yeah that that's exactly right and we're going to need to coordinate and there's this sort of funny thing that happens in Silicon Valley where if you talk to the

engineers making like super intelligence essentially and you and you say is isn't that isn't that impossible and they're like whatever we can do it hold my beer.

Um and if you say okay but to do it safely we're going to have to coordinate. They say don't be delusional.

Yeah, I was been watching you which I find highly uh interesting and and some other people in my research for this thing and it's interesting because you know old stuff checks you know uh like money you know all that stuff is

legislated to high heaven but anything new cryptos is sort of no man's land and we have you know these sort of everything's just legal legal legal legal if it's old and if it's new it's less a fair let's go and it seems that

the newer it is the less regulator or law there is so this is just it's happening in real time AI and it seems like you know we have all kinds of laws for you know intellectual property for for you know people coming

up with this people and then now AI seems to be completely unregulated and and it it sort of it I was thinking about it you know how it is that the newer it is, the the sort of less least

regulated it is and and which leads to the more most amount of problems. Yeah. Yeah. That's that's very well said. Um there's a funny thing that happens with with just humans in our psychology that if something is physical, we treat it seriously. If it's

something is not physical, we don't know how to deal with it. And so, as we've moved from physical systems to digital systems, all of the protections that we've had have gone away. Um, right. If you want to get

somebody physically across the border, a weapon physically across the border into the US, that's a hard thing to do.

If you want to instead like target messages that pit people against each other, um, as our foreign adversaries are doing all the time, there's like actually Facebook's ad and Google's ad, they give you like a white glove service

that says actually tell us exactly who you want to. We can handel the perfect propaganda that pits fellow citizen against fellow fellow citizen. like here's our civil war machine. Please pay us money. Pay us money and we'll we'll

make it happen. Um and that's like so we have a very strong physical border and a non-existent digital border. And the same thing is happening of course with with our with our minds. There's lots of things to protect us if we go out onto

the street. But if we go onto the internet there is really nothing that that protects us. And this is how we've ended up with you know social media which was really just a baby AI sort of like our first contact. Whereas AI in

social media, it's the algorithm that's like ranking uh human content. It's so baby that it can't even generate its own content. It's just choosing which posts get seen and which ones get amplified.

And that tiny misalignment where instead of optimizing for like human flourishing and connection, time spent together, it was optimizing for engagement and time spent on site. That tiny little misalignment was enough to cause like hyperartisanship, polarization, the backsiding of

democracies around the world. Um, the most depressed and anxious generation that we know of. The list goes on and on and on. And now we've moved to like not baby AI, but like true adult monster AI, and it's growing really quickly. And the

question to ask is, have we solved that incentive? And the answer is very clearly no. And I just really want to paint for people um like why why can we say so surely that the incentives point us at an antihuman future. And I think

there's one really good um sort of analogy here which comes from the writer Luke Drago. So he points out the resource curse. What is the resource curse? The resource curse is when a country discovers like a natural resource like oil and suddenly their GDP

becomes dependent on oil. Um their power in the world becomes dependent on oil. um they're not getting much power or wealth from their population. So where do they spend their money? Do they spend their money in oil extraction and

selling infrastructure or into schools and hospitals and welfare programs like obviously into oil? And that's how you end up with like the South Sudans and Venezuelas where you have structural disempowerment and uh mass poverty. And now we're moving from the resource curse into the

intelligence curse where AI is going to give double-digit GDP growth. Like brand new weapons, um brand new uh uh military strategy, um healthcare, like all this stuff is going to be coming mostly from AI, not from humans. So, are governments

incented to put money into you and your future and your kids' future? Um, or into data centers and solar panels? Uh, and obviously it's it's the latter. And actually, you can already hear Sam Alman starting to hint at this.

One of the things that is always unfair in this comparison is people talk about how much energy it takes to train an AI model relative to how much it costs a human to do one inference query. But it

also takes a lot of energy to train a human. It takes like 20 years of life and all of the food you eat during that time before you get smart.

And he's implicitly asking who deserves these scarce resources more. This is how we end up in West Virginia with the cost of electricity being higher than mortgage payments. Like this is how we can tell we're heading into an

anti-human future. And just remember the reason why you and I have power, why people have power is because we uh like corporations depend on us for our labor and governments depend on us for our taxes.

When labor and taxes are coming from AI and not humans, we lose all political uh and economic relevance forever. And then we just have to what? Trust that the AI companies, the ones that took our jobs, are going to be paying our bills

forever. like that does not seem to be a safe world. Which is why, by the way, everyone should go out and vote in these midterms there uh for politicians that are not taking money from the uh the AI super

PACs. There's already been $190 million that gone that has gone into supporting uh essentially anti-human like pro- AAI um politicians. And it doesn't really matter whether they're left or right, Republican or Democrat. Um, and just to put that in context, the last US

presidential election, the most expensive ever was $2 billion was spent. So already 10% of that has been spent just on the midterms alone. So that is the one thing people listening to this can like action out and do right now.

Have you talked to members of Congress and senators directly? A, and if you have, what's their grasp on on all of this because it's highly complex and and then this is a threepart question. And then C, are they able to even like we're

saying, "Hey, something needs to be done. We need to enact laws. We need to regulate this stuff." Are they even able to do that? So, A, have you talked to them? B, do they get it? And C, what's

the plan going forward? Mhm. A, yes, have definitely talked to to to a number of them. B, they are often vaguely aware. um they'll have used like a chat GPT or a Claude, but they don't really uh think deeply about

where is the game theory taking us all. Often they'll get caught in the tropes of but we have to beat China um or like this is good for my my group. Um and see yeah I mean I mean I don't have

the greatest confidence that our current political system is going to be able to like move fast enough but there are places of hope. So if you remember um when Trump was trying to pass the big beautiful bill inside of that there was

a sort of blanket moratorum so that no state could pass any AI regulation or protections for 10 years. Um, and the Senate voted 99 to1 against that. Like when do you ever see that? 99 to1. The the Senate never has that kind of

unonymity. And there's a kind of like um B2B coalition form. call it like the Bannon to uh Bernie coalition because it turns out this is going to be and already is the most unifying issue we have ever seen because where else do you

get like Ralph Nater, Glenn Beck, Bernie Sanders and Bannon all agreeing and in fact in I can't remember exactly and I I can't remember where this stat came from but there's like one of the recent polls I was reading where

you know do you know what percent of the US population is gung-ho like it's like yes it is great to just develop AI as fast as possible no guardrails it it's 5%.

Wow. Like you never get these kinds of numbers. So 95% agree that we're going to need to do something. And so I think this will become the winning political ticket. And that story that I told you about the Alibaba example, if you are a

u a general inside of the CCP as just a human mammal, are you stoked to hear that? No, you're not. And if Trump actually like sat down and listened to this and like and and understood what we're raising toward, would he be

stoked? No. And so at some point we will coordinate not out of like a polyianish sense of like kumbaya coming together, but out of raw self-interest. And our job is to accelerate the speed at which everyone realizes that where we're

heading is anti-human. That is our job. Well, antihuman brings me tuna. So, about 15 years ago, I was talking to Larry Page uh for a long long time, and he was going on I barely understood anything he was saying. Um, and he was going on

about, you know, the synthesis of AI is all human endeavor done by machines and that we're going to need some sort of universal basic income because humans are, you know, and we're going to have to take care of them in some way. And he was

talking about alternative communities and building communities, all this stuff. And I was like, are you what?

Like it was a caricature of of a thug with this sort of Silicon Valley visionary. And um of course 15 years later it all comes true. And basically he was a nerd who's saying if nobody has jobs. I've seen this movie before. The

the thugs come, the currency becomes bullets. they come out of West Virginia and they take all the the the the brainiacs and and and whoever has the biggest hammer wins.

And so he was already thinking about like, you know, how do we how do we sort of get beyond this sort of synthesis of AI, but the synthesis of AI means all human endeavor done by machines. That's the most anti-human thing ever because

work satisfies need, desire, and sanity. And so if you take the jobs away from people, you're taking everything, purpose, you know, community, everything. It's going to be the industrial revolution.

And I always say like, you know, capitalism and communism were just apologizes for a very traumatic time in the industrial. This revolution is 10 times as big. And what happened then is everyone moved from the country to the

cities and a lot of people died. And that was capitalism. And then the communist said, can't we do something a bit nicer? and and now it seems like this is going to be much more traumatic, much more, you know, sort of uh

destratification of of of everything we you know, history and power and all this stuff, Marxist dialectic, all that business.

Uh uh what what happens now? You know, I fear for my kids they're going to get into a job market with no jobs and no universal basic income. there's going to be no solutions because to go back to

our earlier thing of it's so new and there's no regulation that we haven't seen we're not preient enough to say hey let's let's do what Larry Page said humans don't do anything until we have a gun to our heads so we don't have the

solutions we don't have the fixes and and yet we do have the problems so that worries me about governance that's my biggest worry the only guy I talked to was Yang who seemed to even have any idea about what to do about any

of this stuff Yeah. I mean, when in history has one group of people amassed all the money in power and then intentionally redistributed it.

Yeah. Never. Right. Exactly. So, why why do we expect it to go any differently this time?

Yes. I worry. Fill me with fill me with warm and fuzzies. What should we do?

Come on. What then must we do? Yeah. Well, yeah. It it uh is I'm wearing this Kashmir sweater so I I hope that uh that it helps with the warm and fuzzies um is why I do it so I can like

myself soothing. Um I mean there are directionalities for for trail heads and the thing that honestly um gives me any place for non-naive hope is that we haven't really even tried.

Like everyone keeps hiding behind the argument that it's inevitable. And of course when you call it inevitable that's like casting a spell it means that says well if it's inevitable there's nothing for me to do so I don't

do it and so the thing comes true and technological progress may be inevitable but the way we roll it out is not right like we're already seeing in in China court cases that say that companies are not allowed to like fire workers to

replace them with AI. Um I'm not saying you know the US should do what China does but there China is moving down a path to distribute the technology wisely within its borders at least more with more guard rails than than the US. Um so

everyone just hides behind the inevitability thing. What percentage of time have all those like billionaires spent on actually trying to coordinate?

Have they spent a month doing it? Yeah, exactly. uh what percentage of their wealth have they spent on like coordination after human beings invented the atomic bomb?

We had to invent a whole bunch of stuff. We had to invent like seismic monitoring and satellite imaging and then everyone came together was forced to come to Brettonwoods sit in a hotel for something like six weeks 100 delegates

from 100 nations to figure out what do we do about this? Have we done that yet?

No. Um and so I think there's a big you know uh having tried it overhang we're just we have not taken seriously the risk enough to say let's get together and yeah the UN is not like defunct we

don't have great like coordination bodies but this is the moment to figure it out and that'll start to happen as we get clarity about the nature of the problem and you you saw it after mythos came out with the ability to hack almost

any computer on Suddenly uh it got added to the agenda for Xi and Trump to sit down and start talking about it. And while we don't really know what happened, we know they at least expressed a willingness to

think about coordination, which is much better than we were before where they're like absolutely not. This is this is not something we're going to talk about.

So I don't think that'll fill you with hope per se, but at least maybe like a shinering direction. And it's really hard because just to say there is no linear path. There's no story I can tell you that'll actually get us from here to

there. This is too complex with the largest set of incentives and trillions of dollars and permanent dominance as the prize at the end of the rainbow.

This is the hardest coordination problem humanity has ever had to face by far. But that doesn't mean that it's impossible. It just means it's very very very very difficult. And that gap is where I still try to find uh some kind

of hope. Um good because I'm going to ratchet it up in my next question.

Okay. You mentioned China and America. We've talked about disproportionate wealth. Um very few people owning the technology and the the wealth created by that technology. We've talked about the synthesis of of AI, meaning nobody has jobs anymore. Um, we've talked about um,

you know, drones, AI and warfare, Terminator dystopian type futures. We haven't even gotten into the marriage. And you're a physicist, so you're going to love this between quantum, which is now coming online, and AI quantum. And if you look at quantum

and you want to talk about wealth and technology, I mean, it's basically a two-horse race between America and China. um you know it's proving sort of quantum physics by its existence you know the quantum computing and and you have that's going to be a sea change

in computing power that builds a whole like you were saying well now everybody now it's like the Macintosh everybody had to have a personal computer now that everyone has to have a quantum that's populated by AI. This it's going to be a whole new

technology that leaves everybody in the dust. So, it's another ratchet to what we're already talking about and it's brand new and we certainly don't know how to regulate that.

And so, you now have AI and quantum being a sort of doublepronged double whammy we call that in the technical term. And and it's two countries sort of leading the thing. And by the way, everyone else is being left in the dust.

Yeah. Yeah. So sometimes I actually think what we're racing towards is a mutually assured political revolution because wow um because and actually China in some sense is even more exposed than the US for the number of people displaced. As

the number of people displaced and their jobs go away, their ability to make livelihoods go away, the pitchforks come out. And so it's sort of like asking for like regime change by going as fast as possible. And it's in neither country's

best interest to do that. Yes. Yeah. also more fewer and fewer and fewer people get more and more money as you're looking.

That's right. And note that before the um like the the the sort of the elite class whoever like and actually it's not even elite class, it's going to be like eight people um when those eight people would rely for their safety on like private

armies and those private armies or mercenaries are humans um which can defect. But that's not going to be the case um with robotic armies. And so it's just really important to note that our dynamics are very different. The things

that enabled us to be able to revolt and fight back like that has fundamentally changed in the age of like drone and robotic armies. Um and so we really do have this very brief window. And what you're pointing out is you know we're

starting to get to quantum a new advance in uh computing. Um what we should expect to see what Dario Amade from Anthropic says we'll get a century's worth of technological advancement in a single decade. We're going to see those

things happen faster and faster and faster and faster. Um and as Connor Leehy says in the AI doc the the film that we've worked for the last two and a half years with the directors of everything if we're all at once to like

get out to like help everyone see this thing. He says, "Right now, there is more regulation on making a sandwich in New York City than there is on making potentially like humanity ending AI." And we're going to have to like bound

that responsibility or bound that power with responsibility. And the reason why I went to the mutually assured political revolution is obviously coordination is going to become necessary at some point. And our choice is either we wait until you know

a very catastrophic thing happens like major infrastructure goes down. We have to shut off power in the internet to try to contain it or we can start to get our act together now. And it's way less pain now. And you could imagine there being

some kind of, you know, limit on compute or just like a token tax or a CPU or like a GPU tax um that the China and the US says, hey, it's in all of our interests not to get these crazy

self-improving models that run away um and do terrible things. Let's at least decide between stuffs where the sensible limbs are going to be and take like steps forward into the future because it's in everyone's self-interest.

Now to add another spanner into the works, yeah, you've got China and America, quantum AI zooming off into the future. And if you're someone like Russia, you say, well, Elon's, you know, SpaceX is, you know, the biggest IPO, but they

also control, you know, all the data that's coming in and and and pretty a lot of the computes going to be done up there and, you know, emails, internet, banking, etc. So, this tremendous amount of power is passing you by. You used to

be the second superpower. Maybe you still are. And you say, "I have these analog nukes over here that that are my hammer." To go back to Larry Paige, I've got my hammer over here. You guys have your big brains.

It's sort of like Planet of the Apes. We're going to come and you guys have your your your tricks and your things, but we're going to come with our guns.

Um, that's another terrifying thing. You know, we're fighting an asymmetric uh war now with Iran and they have their Shahed drones that are going up against all of our technology. It's always unforeseen, but there's going to be some

left behinds that are going to be hugely problematic because they're going to say, "Hold on a second.

You're not going to leave us eating mud while you're, you know, up in space eating steak." Like, what uh what what do what do you think happens there?

Yeah. I mean, this is a different form of the Truscidities trap where like before you lose your power forever, use what you got to keep the other guy from dominating you permanently. And that's exactly what what you're naming. Um, and

actually a lot of the psychology of the people running the labs isn't even so much now like I need to win, it's I can't lose to you. like um so this is why you know a number of people believe that it's not actually a

race to AGI it's a race to World War II because whoever gets that permanent strategic advantage like the mythos that can hack like at the next level up like any computer get into any nuclear power plant can shut down like the opponent's

weapons like whoever gets that first has permanent dominance and no country will let the other country have that. China won't let the US have that. the US won't let China have that. Um, and this is there's there two branches here. There's

the super scary and and that's how we end up with like kinetic strikes. Um, like maybe it's not so wise for me to be living near San Francisco. It's sort of like a major hub. um or people see like in leaders better

than all out war and heading into World War II is we do the coordination now and we agree that no one player can get access to like that kind of permanent dominance. Um, I don't know if that's going to happen,

but this is why I keep coming back to the confusion in people's minds that when they say we have to beat that, when we have to get that strategic dominance forever, they are still imagining the thing they control,

right? It happens. It's so subtle. It just happens in their mind as soon as and then you flip them back into but that's actually a sociopathic genius doesn't care about you and is banded together with other AIs. Um, then you

realize that we're that we're racing towards the wrong thing. Which is why I think we have to redefine the nature of the race that we're in to who can make the strongest cybernetic society that is good at governing. To put it just really

simply, what happens when you accelerate and you don't steer? You crash. So this shouldn't be a race to just accelerate. Should be a race to steering. Like that's how you actually Yeah. And I agree. And on that positive

note, we're going to do some warm and fuzzies. Oh, good. When I gradu when I graduated college, I started working for Greenpeace because I was like, well, it's politically everything's nuanced and confusing, but environment good, people bad. You know what we do to the

environment, so I'll go work for Greenpeace. And then Greenpeace just ate itself because people can't can't do anything.

Um, but your your Earth species project. Now, I love this because it's sort of communicating and you again, correct me if I'm wrong, but sort of using AI to communicate uh with animals and you found all kinds of

amazing things, but I want you to tell me the warm and fuzzy uh stories there, whales, uh you know, dolphins. I think there's one where a chicken makes a certain sound to the to the to the baby so they have names and all this stuff.

Fascinating. Maybe just let's talk about if we do get it right and we're not we're we're not Star Wars, we're Star Trek. It's not dystopian. It's sort of moderately utopian and we get it right and we have we don't have to work. So it

frees us up to do things like communicate with our nature and our world and animals and trees and god knows what else. Give us I mean I'm I love this. I do not know anything about it and I'm

pretty sure a lot most people don't either. So give us give us the sort of the the primer on that.

Yeah. Yeah. So this is Earth Species Project and the the whole goal of it is if we can understand the language of the other species of Earth and the other cultures of Earth.

Um well it changes how we relate to the rest of nature. Like it opens us up to the awe and the wonder um of of what's what's around us. And so just to give a couple examples, um, uh, it's, you know,

so it turns out parrot parents will spend the first couple years or a couple weeks of their chick's life leaned over each chick whispering a unique name which they will then use for the rest of their life.

Amazing. Which is just so so beautiful. Like what's what's deeper than having a name and an identity? And it turns out dolphins do the same thing. Elephants do the same thing. Um, Dolphins, it turns out, um, not only will like use a name

that their mothers give them, but they can refer to each other. They'll talk about another dolphin that isn't there by name. Um, uh, and one of the researchers that just joined our team just before she joined Earth Species,

you know, she discovered that they will continue to use their mother's name even after she's passed away.

Wow. which again so profound because you're talking I mean we're talking about like like lineage and uh and and memory and family and we don't know what they're saying but you can imagine relationships exactly uh legacy it's it is so it's so

profound um and you know we can just keep going another another crazy dolphin experiment just I think is very interesting is 1994 they um they gave dolphins they sort of were teaching dolphin parents different gestures and the first gesture was do

something you've never done before. Um, and they'll actually learn how to do that. They will learn how to um well understand that concept, remember everything they've done before that session and invent a new thing that they've never done before, which is

already cool because innovation is something we think is probably just humans. Um, and then they'll give a second gesture do something together.

And they say to two different dolphin pairs, do something you've never done before together. and they go down and they exchange sonic information and they come up and they do the same thing they've never done before at the same

time which just blows my mind. And the thing that we keep discovering is, you know, just like when we pointed the Hubble telescope back in 1995 um or we scientists did at the at an empty patch in sky

and what did they discovered? They didn't discover nothing. They discovered the most number of galaxies we had ever seen before in a single spot. And that's what we're discovering as we're pointing our new AI tools at the natural world.

What we're discovering is that actually we know almost nothing. And to give an example from our own research um with crows like we think we all know crows.

We hear them all the time. What we found out is that the sum total of what humanity knows about crows that has been published on that scientists talk about was only 30% of their actual communication system. So the super

majority 70% of crow communication turns out are these quiet close intimate calls sort of whispers they make while they fly. Um and science is just completely unaware of it. So we really believe in the next you know four or five years

we'll be able to deeply understand and peer into the other minds and cultures of earth and then our hope is that there's a kind of my octopus teacher moment where you know when I talk to people about the film my octopus teacher

asked how many people have watched it many people raised the hand how many people stopped eating octopus many of those hands stay up and is that because the movie made you feel bad was there shame guilt involved no it just opened

the door to the wonder and love and awe of what was already there. Um, and this is how we then end up at that line. The way we treat nature is the way we the AI will treat us. We're just following our

incentives right there. Yeah. Yeah. I have that as my next thing right there. I mean, you you explained it beautifully. I want to I want to just get two like what species has because this is I find this a fascinating and b

as you're talking I'm like maybe we're not doomed. Um uh uh what what species uh is has surprised you the most so far?

Like what what are the biggest sort of wow, you know, kind of Yeah. Well, I mean the the one I just gave on crows just blew my mind. Um crows.

Uh the other one is beluga. Beluga are insane. When you actually listen to beluga, they don't sound like a dolphin or humpback. They have this thing that sounds almost like it's like They sound like an alien modem. They

speak, it seems, more in like modem like packets. Um, when they give their hells and there you're like, okay, so 70% was pretty bad for science to be unaware of for crows. For belugas, it's 97%.

Um, amazing. Uh, for dolphins, like half of the communication science says is um is signature whistles or names. um even though there's a lot of variations that they don't understand but okay let's just assume half communication is names

the other half they just call non-signature whistles. So it's sort of like you're trying to understand human communication and you're like okay we've got it figured out human communication consists of names and then other things they say which are non-names like that's

where we are. We're just I always there's a graph that I love to show which is um I call it the most humble graph because it includes all of time and all of space um and it's just showing what is the range of the human

experience and if you try to draw the box of what can be perceived to what human beings perceive it's a dot so small you can't see it like the sum total of that which is out there to be experienced is

unexperiencable by us and that vastness and wonder I think is only accessible to us via the empathy these other cultures.

That's amazing. It's like the Enigma when the British captured the Enigma, but they still had to figure it out and they they had they figured out the date and they figured out at the end it said Hy Hitler. And so those letters those

letters and the date they could then go reverse and go back into the encryption.

And and so we're at that sort of first stage of, you know, we're just we got names and no names. But when that no names becomes everything else, the actual meat of the message, then and it's going to be really uh um

mind-blowing. The the there's an AP piece and whenever I look down, it's not because you're not scintillating, it's because I'm trying to keep up with your large brain. the a there's an AP piece that says, "Your real goal isn't a a

sci-fi animal translator, but rebuilding humanity's relationship with nature." That's right. And I thought that was I thought that was great. Um, you know, is there a path towards sanity where we get it right with AI, we get it right with quantum,

we we realize that mutually assured destruction or mutually assured revolution or mutually assured like any of these in insert dystopian end here.

Um, and we say, "Hey, let's use this technology to get in touch with space, to get in touch with more galaxies, to get in touch with uh dolphins and and ravens and crows and and other things that we can't even, you know, begin to

imagine like trees and what have you." Um, is there a path forward to to that?

Well, as I said before, there's not a linear path. There's no easy way for me to chart this is going to happen, then this is going to happen, then this is going to happen.

Um, and so then this really becomes a question not just of like what we must do, but the who we must be as a species.

We are undoubtedly in our technological adolescence. And if we are not the kinds of people who look for the path, then if there were to be a path, we definitely wouldn't see it. We definitely wouldn't find it. So

this is now about the conviction of trying to look for the path with more and more and more people and another way of saying this is you know human beings have always had a relationship with technology and generally speaking we'll encounter

some problem and then we'll build a technology to kick the can of that problem down the road but in the process we make more and bigger problem right like human beings we have too many human beings um on the planet we can't feed

them so we uh fertilizer, but fertilizer creates nitrogen runoff and dead zones in the ocean, all these other things. Um top soil like devastation. And so now they're like even bigger problems. So like industrial era technology makes super industrial era problems that we

need to solve. So we kick the can down the road and our choice is either we coordinate or invent the next era of technology AI uh to solve those problems like climate change and whatnot. But in the process, we're going to invent even

bigger like super AI scale problems. Um, so now the problems are as big or bigger than the world. Forever chemicals everywhere, AI, everything that we're talking about now.

Plastic. Plastics. Exactly. And all of our bodies. And if we miss kicking a can even once, that's sort of it. Um and this is why this is our right of passage where either we go through and face death and in the process mature

or we don't make it and those are the two options that I see and how we show up in the next 24 months to couple years will determine that path and so the hope is is that if things are clear enough if

we're not sleepwalking into it then at least At least we have choice and at least we can't say we didn't know. I know that's not like the most hopeful version of this answer, but I think it's the accurate version.

We've talked about AI. We've talked about quantum. We've talked about the beautiful talking to the animals, which I love.

I want to get into the big metaphysical stuff. Now, talked about the Hubble telescope.

Um, dark matter. I love dark matter because I love this. You know, there's too much gravity. the planet should be spinning off into space.

Dark dark matter, wimps, you know, the fifth dimension, all of these delightful things. Um, what fascinates you about dark matter? A and how do you think it ties into all of this, you know, that's going on today with quantum and and and AI and h like

it seems to me that we're going to have this sort of consciousness mindblowing sort of, you know, Moors law on steroids happen over the next 10 to 15 years.

Yeah. Yeah, that's right. And just so people can hear it at home, inside of the AI labs, what we're hearing from the people that we know is that they're living inside the singularity where human beings are not really writing

Claude's code anymore. Claude is writing its own code, which means we are at the beginning of that recursive self-improvement where it makes AI makes its own smarter successor, making its own smarter successor faster and faster and faster with none of that code having

been written or even really reviewed by humans, which is like that is the nightmare scenario. And actually, if you want to make it even worse, the nightmare scenario is when that's happening within private companies all competing with led by billionaires that

hate each other. Like that is the worst possible world. Exactly. That that is Yeah, that is fake. Like Sam Alman and and Dario won't even hold hands.

Um Yeah. And Elon is in there too, right? Wait, but what was your question?

I just I totally riffed myself into a No, I love the riff. I was just saying you're a smart guy. It seems to me that quantum AI these things that are happening you which you put eloquently just now but if you add the things that

we're finding out i.e. dark matter, wimps, dark energy. Um it seems like we're going to have a mindblowing amount of um information that comes like for example quantum you know quantum physics used to be like what you know and then

quantum the first and we had the Google guy the head of quantum at Google on and he's like yeah the first quantum solution was that quantum physics is right and that you know there's infinite possibilities and infinite timelines and

infinite universes and you're like what's happening? So what do you think over the next few years when quantum comes online and AI is really cranking and like it's Moors law was sort of like in your era and my era the Apple era was

like okay every few years we get a thing and every oh now we have an iPod okay now now we have an iPhone you know now it seems to be going just on steroids and I I mean guys like you what do you

see happening and and and Maybe I'll make it more concrete. What's one great thing that you see and what's one terrible thing that you see?

I guess the to start by saying is both the optimists and the pessimists don't go far enough because we are deeply limited by our imagination. So all the amazing things AIs are going to do, whether it's like, you know, solving cancer or new

antibiotics, helping us speak to animals, um or rather to learn how to listen to them, um like all of those are just the like the very obvious first things that AI could possibly do. And AI is hacking critical

infrastructure, taking over nukes, persuading people, doing uh psychological uh breakdowns for an entire population.

These are and biological weapons. These are also the very obvious first things that we think of. And if we go back to the invention of the transistor, which of course led to the entire computing revolution, when you looked at what they

did, they were imagining that these transistors could maybe help do better telephone switches and maybe make hearing aids. They couldn't have possibly imagined how far we've come.

And so I think the highs are going to be much higher and also the lows are going to be much like the catastrophes are going to be much much much worse. And it's just important for people to realize that there's an asymmetry here

where that is the good sides don't preclude the downsides but the downsides do preclude the the good sides that is even if we get cancer solving drugs if they land in a world that has been broken by AI epidemics it doesn't

matter. So I just want people to hold that. Um yeah, in terms of like what's one really good thing, I don't know if this is good or bad but you know when when I stay up to date with all of the the latest in particle

physics and um you know there's amazing researchers like Nema Connie Hamemed who's doing work down to the very fundamentals of particle scattering discovering that perhaps space and time are emergent behaviors or something lower more fundamental um uh there is

some very interesting work that's happening at the synthesis of mathematics and physics which is to say that the deeper in physics you go the more fundamental in mathematics you go and the question I think in a subset of

um both physics and mathematicians mind is are they actually the same that is why is it that when you go off and you invent very weird mathematics even imaginary numbers the square root of negative 1 what does that even mean

Somehow it really well describes the world. This is the unreasonable effectiveness of mathematics to to quote the famous paper. What happens if actually the structure of the universe is actually the structure of mathematics that they're the same? And there's a lot

of very compelling evidence that says well maybe they are. And that's I don't want people to imagine like that means the symbols that we do mathematics with.

That's not what I'm talking about. The symbols are pointing at a structure. And if that's true, that means anything that happens in the digital domain in symbols, AI will be able to solve.

That's why anything with theoretical in front of its name, theoretical physics, mathematics, um that gets solved by AI much faster than anything that requires going back out to the physical world because it takes actually testing the real world to to understand. And so if I

go out on a live I don't know in like 10 15 years maybe we discover that actually math and physics are the same in which case the things I will discover are going to be the most insane and I I

cannot possibly speculate. Well on that note that was just fascinating as Rascin uh incredible to spend time with you.

Let's keep the lines of communication open because uh it was pretty mind-blowing. I liked it all. Uh, thanks for taking the time, man.

Yeah, Shen, thank thanks so much. Thanks so much for going there. As I said, I know this is not easy material. And the one thing I just leave listeners with besides you should go watch the AI doc and you should listen to your undivided

attention is when you hear all of this, it can sound very disempowering, very demotivating. And there's sort of two places the mind goes. Either you like you take it all in and you say, "It's too much. I'm now depressed." Um, or you

say, "Well, that can't all be true. I'm like, actually, when I look around the world, it's not so bad." And so, I'm I'm just going to ignore I'm just going to ignore it. I can't actually write. I'm going to go back and I'm going to use

the blinking cursor chat PT. It's just helping me. Where's the existential threat? I forgot.

I just want everyone to know it is not on your shoulders to solve the whole problem. The most important thing for most people to do is just to break the spell of inevitability whenever people bring it up and say actually there is

another path be part of the collective immune system because it is that shared knowledge when I know that you know that I know that let something else happen you know and my moment of hope here is if we don't just call this an AI problem

but we call this a technology encroaching into our humanity problem then actually we have way more momentum than we ever thought possible even 2 years ago. Like following the social dilemma, eventually Australia banned social media for teens under 16 because

Australia took a step forward and became the moral leader articulating something that everyone feels. they became the first suddenly you know like Spain, the UK Denmark France Malaysia Indonesia and now India I think have all either announced that they have or will ban

social media for teens and that means as of today 25% of the world's population lives in such a state where they're protecting kids from malescentized technology and that is the human movement that's underway. they cause like the as people lose jobs the human

movement is just going to go stronger and stronger and the only question is as billions of people like end up in the human movement does it happen in time for us to actually shift the political headwinds so that's I want to leave

everyone with that's a good one the human movement I love it Asa Raskin thank you so much for your time thank you so much Shane it's been real pleasure we use PDF spaces to get deep into the documents ments that I relied on to

research this episode. You can interact with them, too, on this episode's PDF space linked in the description.