Skip to content
The Joe Rogan ExperienceThe Joe Rogan Experience

Joe Rogan Experience #2311 - Jeremie & Edouard Harris

Jeremie Harris is the CEO and Edouard Harris the CTO of Gladstone AI, a company dedicated to promoting the responsible development and adoption of artificial intelligence. https://superintelligence.gladstone.ai/

Joe RoganhostJeremie HarrisguestEdouard Harrisguest
Apr 25, 20252h 47mWatch on YouTube ↗

EVERY SPOKEN WORD

  1. 0:00 – 0:41

    AI “Doomsday Clock” framing and why the timeline feels compressed

    1. JR

      (drum music) Joe Rogan podcast, check it out.

    2. NA

      The Joe Rogan Experience.

    3. JR

      Train by day, Joe Rogan podcast by night, all day. (rock music) All right, so if there's a Doomsday Clock for AI, and we're- we're-we're fucked, what- what- what time is it?

    4. JH

      Oh, wow. We're-

    5. JR

      If midnight is, we're fucked.

    6. JH

      We're getting-

    7. EH

      Getting right into it.

    8. JH

      You're- you're not even gonna ask us what we had for breakfast, like what-

    9. JR

      No, no, no, no, no, no, no. (laughs)

    10. JH

      (laughs) Jesus. Okay.

    11. EH

      Let's- (laughs)

    12. JH

      (laughs)

    13. EH

      ... let's get freaked out.

    14. JH

      Well, okay, so- so there's one, um, without speaking, like, (laughs) the- the fucking Doomsday dimension right out the gate-

    15. JR

      (laughs)

    16. JH

      ... there's a question about, like, where are we at in terms of AI capabilities right now, and what do those timelines look like?

    17. JR

      Right.

  2. 0:41 – 2:23

    Task-horizon benchmarks: measuring agent capability and extrapolating to 2027

    1. JH

      There's a bunch of disagreement. Um, one of the most concrete pieces of evidence that we have recently came out of a- a lab, an- an AI kind of evaluation lab called Meter. And they put together this- this test. Basically, it's like, you ask the question, um, pick a task that takes a certain amount of time, like an hour, that takes, like, a human a certain amount of time. And then see, like, how likely, uh, the best AI system is to solve for that task. Then try a longer task. See, like, a 10-hour task, can it do that one? And so right now, what they're finding is, um, when it comes to AI research itself, so basically, like, automate the work of an AI researcher, you're hitting 50% success rates for these AI systems for tasks that take an hour long. And that is doubling every, right now, it's, like, every four months.

    2. EH

      S- so-

    3. JR

      Hm.

    4. EH

      ... like, you had tasks that you could do, you know, a person does in five minutes, like, you know, uh, ordering an Uber Eats or, like, something that takes, like, 15 minutes, like maybe booking a flight or something like that. And it's a question of, like, how much can these AI agents do, right? Like from five minutes to 15 minutes to 30 minutes. And in some of these spaces, like research, software engineering.

    5. JR

      Mm-hmm.

    6. EH

      And it's getting further and further and further and doubling, it looks like, every four months. So it's like-

    7. JH

      If you- if you-

    8. EH

      Yeah.

    9. JH

      ... extrapolate that, uh, you basically get to tasks that take a month to complete, like by 2027, tasks that take an AI researcher a month to complete, these systems will be completing with like a 50% success rate. That's where this goes.

    10. EH

      So you'll be able to have an AI on your show and ask it what the Doomsday Clock is like by then.

    11. JR

      Uh, it probably won't laugh. (laughs)

    12. JH

      (laughs)

    13. EH

      (laughs) That's gonna be part of the problem.

    14. JH

      It'll have a terrible sense of humor about it by then.

    15. EH

      Just make sure you ask it what it had for breakfast before you start, I guess, but-

    16. JR

      Mm-hmm. (laughs)

  3. 2:23 – 3:25

    Quantum computing and AI: real advantages, wrong assumptions, and near-term irrelevance

    1. EH

      Yeah. Um, what about quantum computing getting involved in AI?

    2. JH

      It's... So, yeah, I, honestly, I don't think it's... I- if you think that you're gonna hit, uh, human-level AI capabilities across the board, say 2027, 2028, which when you talk to some of these, the people in the labs themselves, that's the timelines they're looking at. They're not confident, they're not sure, but that seems pretty plausible. Uh, if that happens, really, there's no way we're gonna have quantum computing that's gonna be giving enough of a bump to these techniques. You're gonna have standard classical computing. Um, one way to think about this is that the data centers that are being built today are being thought of literally as the data centers that are going to house, like, the artificial brain that powers superintelligence, human-level AI when it's built in, like, 2027, something like that.

    3. JR

      Hm. So how would... K- how knowledgeable are you when it comes to quantum computing?

    4. JH

      So, uh, d- a little bit. I mean, I, like, I did my, um, my grad studies in, like, the foundations of quantum mechanics.

    5. JR

      Oh, great.

    6. JH

      Um, yeah, well, (laughs) it was a mistake, but I appreciate it for the person

    7. JR

      (laughs)

    8. JH

      ... the person...

  4. 3:25 – 17:18

    Academia’s incentive traps vs startups: credit wars, gatekeeping, and reality checks

    1. JR

      Why was it a mistake?

    2. JH

      You know, so academia is this f- like, kind of funny thing. Um, it's really bad culture. It- it teaches you some really terrible habits. So, basically, my entire life after academia, and Ed's too-

    3. JR

      Yeah.

    4. JH

      ... was unlearning these, like, terrible habits of... It's- it's all zero sum, basically. It's not like when you're working in startups. It's not like, you know, when you're working in tech where you- you build something and some- somebody else builds something that's complementary, and you can team up and just, like, make something amazing. It's always wars over who gets credit, who gets their name on the paper. Did you cite this fucking stupid paper from two years ago 'cause the author has an ego and you got to be on... I was literally, at one point, um, the, I- I'm not- (laughs) I'm not gonna give you any details here, but, like, there was a collaboration that we ran, uh, with, like, this, anyway, fairly well-known guy. And my supervisor had me, like, write the emails that he would send from his account so that he was seen as, like, the guy who was, like, interacting with this bigwig. K- uh, that kind of thing is, like, d- doesn't tend to happen in startups, at least not in the same way, 'cause everybody's-

    5. JR

      So he wanted credit for the, like, he wanted-

    6. JH

      For just-

    7. JR

      ... to seem like he was the genius who was facilitating this?

    8. JH

      Yeah.

    9. EH

      For sounding smart on email.

    10. JR

      Oh.

    11. EH

      Right?

    12. JH

      But- but that- that- that happens-

    13. EH

      That- that happens everywhere.

    14. JH

      Dude-

    15. EH

      And-

    16. JH

      Yeah.

    17. EH

      ... the reason it happens is that these guys who are, like, professors or even not even professors, just like your postdoctoral guy who's, like, supervising you, they can write your letters of reference and control your career after that- that lab-

    18. JH

      Yeah, they got you by the balls, like-

    19. EH

      ... so they can do whatever. They can do whatever. And so what Jer did-

    20. JR

      Oh, God, it's like a movie.

    21. EH

      Totally.

    22. JH

      Yeah, it's gross.

    23. JR

      It's like a gross movie.

    24. EH

      It's disgusting.

    25. JH

      It's ab-

    26. JR

      It's like a gross boss in a movie that wants-

    27. EH

      (laughs)

    28. JR

      ... to take credit for your work.

    29. EH

      And it's real. It's rampant. And-

    30. JR

      Ugh.

  5. 17:18 – 21:35

    Machines building machines: industrial-revolution déjà vu and AI’s physical footprint

    1. JH

      He also wasn't even the first person to come up with this idea of machines building machines and there being implications like human disempowerment. So if you go back to ... I think it was like the late 1800s, and I- I don't remember the guy's name. But he sort of like came up with this ... He was observing the Industrial Revolution and the mechanization of labor, and kind of starting to see ... More and more, like if you zoom out, it- it's almost like you have an- humans are an ant colony, and the artifacts that that colony is producing that are really interesting are these machines.

    2. JR

      Yeah.

    3. JH

      You know? You kind of like look at the surface of the Earth as this like gradually increasingly mechanized thing, and it's not super clear, if you zoom out enough, like what is actually running the show here?

    4. JR

      Right.

    5. JH

      Like, you've got humans servicing machines, humans looking to improve the capability of these machines-

    6. JR

      Right.

    7. JH

      ... at this frantic pace. Like, they're not even-

    8. JR

      Right.

    9. JH

      ... in control of what they're doing.

    10. JR

      Are we the-

    11. JH

      Economic forces are pushing-

    12. EH

      Are we the servant or the master, right?

    13. JH

      Yeah.

    14. EH

      At a certain point. Like, yeah. Y- y- and the whole thing is like especially with the competition that's going on, um, between the labs, but just kind of in general, you're at a point where like ...... do the CEOs of the labs, like, they're, they're these big figureheads. They, they go on interviews, they talk about what they're doing and stuff. Do they really have control over the, any part of the system?

    15. JR

      Oh, yeah.

    16. JH

      The economy's in this, like, almost convulsive fit, right? Like, you can almost feel like it's like it's hurling out AGI.

    17. EH

      (laughs)

    18. JR

      Mm-hmm. Yeah.

    19. JH

      Um, and, like, like, for as one kind of, uh, I guess, uh, data point here, like, all these labs, so OpenAI, uh, Microsoft, uh, Google, every year they're spending like an aircraft carrier worth of capital, individually, each of them, just to build bigger data centers, to house more AI chips, to train bigger, more powerful models. And that's like ... So, so we're actually getting to the point where if you look at on a, a power consumption basis, like, we're getting to, you know, two, three, four, 5% of US power production, um, i- if you project out into the late, uh, 2020s, um, kind of-

    20. EH

      You mean 2026, '27, you're talking about.

    21. JH

      Uh, well, n- not for double digit, though.

    22. EH

      Not for double digit.

    23. JH

      Yeah.

    24. EH

      But for single digit, yeah, you're talking like that's a few gigawatts. So one gigawatt is-

    25. JH

      So, so n- uh, not for single digit. It's in the, the, like... For, for 2027, you're looking at like, you know, in the .5-ish percent. But it's like, it's a big fucking fr... Like, you're talking about gigawatts and gigawatts. One gigawatt is a million homes. So you're seeing like one data center in 2027 is easily gonna break a gig. There's gonna be multiple like that. And so it's like a thousand... Sorry, a million-home city, metropolis really, that is just dedicated to training like one fucking model.

    26. JR

      Whoa.

    27. JH

      Like that, that's what this is. Again, if you zoom out at planet Earth, you can interpret it-

    28. JR

      (clears throat)

    29. JH

      ... as like this, like all these humans frantically running around like ants just like building this like artificial brain. It's like one super mind-

    30. JR

      Yes.

  6. 21:35 – 25:10

    Salt Typhoon and telecom backdoors: how lawful access becomes foreign access

    1. JR

      What's Salt Typhoon?

    2. JH

      Um, so Salt Ty... Oh, sorry. Yeah, yeah. So it's this big, um, Chinese cyberattack. Actually, this starts to get us to, uh, to kind of the, the broader-

    3. JR

      That's such a great name, by the way, Salt Typhoon.

    4. JH

      You know-

    5. JR

      Fuck yeah, guys.

    6. JH

      I really wish I would...

    7. EH

      (laughs) That's like the coolest name. They have the, they have the coolest names for their cyber operations meant to destroy us.

    8. JR

      That's a great name. Yeah, Salt Typhoon is pretty slick.

    9. JH

      They got a st- You know what it's, it's kind of like when, um, when people go out and do like a, a- an awful thing, like a school shooting or something, and then like, "Oh, let's talk about..." You know, if you give it a cool name, like now the Chinese are definitely gonna do it again. Um, anyway, that's-

    10. JR

      'Cause they have a cool name?

    11. JH

      Yeah, that's definitely a factor.

    12. JR

      Salt Typhoon.

    13. EH

      Salt Typhoon.

    14. JR

      Pretty dope.

    15. EH

      Dope. Yeah.

    16. JH

      But it's this thing where basically... So, so there was in the, um, the 3G kind of protocol that was set up years ago, um, law enforcement agencies included backdoors intentionally to be able to access comms, you know, theoretically-

    17. EH

      Mm-hmm.

    18. JH

      ... if they got a warrant and so on. And, um, well, you introduce a backdoor, you have adversaries like China who are wicked good at cyber, um, they're gonna find and exploit those backdoors. And now basically they're, they're sitting there, and they had been for some people think like maybe a year or two before it was really discovered.

    19. JR

      Hmm.

    20. JH

      And just a couple months ago, they kind of go like, "Oh, cool. Like, we got fucking like China all up in our shit." Um-

    21. EH

      And this is like, this is like flip a switch for them and like you turn off the power or water to a state. Or like you fucking... Yeah. You know.

    22. JH

      Well, sorry, this is, sorry. Salt Typhoon though is about, um, just, uh, sitting on the, the, like basically telecoms network.

    23. EH

      Oh, that's the telecom one. That's right.

    24. JH

      Yeah, yeah, yeah. It's not the... But, but yeah, I mean, that, that's another thing.

    25. EH

      There's another, there's another thing where they're doing that too.

    26. JH

      Yeah. And, and so this is kind of where... What, what we've been looking into over the last year is this question of how, what is... If, if you're gonna make like a, a Manhattan project for superintelligence, right, which is, that's, I mean, that's what we were texting about, like way back and-

    27. EH

      Right.

    28. JH

      ... and then, um, actually, funnily enough, we, we shifted, right, our date for security reasons. But, um, if you're gonna do a Manhattan project for, for superintelligence, um, what does that have to look like? What does the security game have to look like to actually make it so that China's not all, all up in your shit? Like today, it is extremely clear that at the world's top AI labs, like all that shit is being stolen. Like there, there is not a single lab right now that isn't being spied on successfully based on everything we've seen, um, by the Chinese.

    29. JR

      Can I ask you this?

    30. JH

      Yeah.

  7. 25:10 – 37:55

    Espionage reality check: ‘The Thing,’ passive bugs, and extreme willingness to harm

    1. EH

      So, I'll gi- I'll give an example. Like, I'll- I'll tell you a story that's- that's kind of like, it's- it's a public story, and it's from a long time ago, but it kind of gives a flavor of, like, how far these countries will actually go when they're playing the game for fucking real. So, it's 1945. America and the Soviet Union are like best pals because they've just defeated the Nazis, right? To celebrate that victory and the coming new world order that's going to be great for everybody, the children of the Soviet Union give as a gift to the American ambassador in Moscow this beautifully carved wooden seal of the United States of America. Beautiful thing. The ambassador's thrilled with it. He hangs it up on, behind his desk in his private office. You can see where I'm going with this probably.

    2. JR

      Oh, yeah.

    3. EH

      But yeah. Seven years later, 1952, it finally occurs to us, like, "Let's take it down and actually examine this." So they dig into it, and they find this incredible contraption in it called a cavity resonator. And this device doesn't have a power source, doesn't have a battery, which means when you're sweeping the office for bugs, you're not going to find it. What it does instead is it's designed... That's it. That's it. It's-

    4. JH

      The thing.

    5. EH

      It's beautiful. They call it-

    6. JH

      They call it the thing.

    7. EH

      They call it the thing. And what this cavity resonator does is it's basically designed to reflect radio radiation back to a receiver to listen to all the noises and conversations and talking in the ambassador's private office. And so-

    8. JR

      How's it doing it without a power source?

    9. EH

      So, that's what they do. So, the Soviets, for seven years, parked a van across the street from the embassy-

    10. JR

      (laughs)

    11. EH

      ... had a giant fucking microwave antenna aimed right at the ambassador's office, and were, like, zapping it and- and looking back at the reflection and literally listening to every single thing he was saying. And the best part was, when the embassy staff was like, "We're going to go and, like, sweep the office for bugs periodically," they'd be like, "Hey, Mr. Ambassador, we're about to sweep your office for bugs." And the ambassador was like, "Cool. Please proceed and go and sweep my office for bugs." And the KGB dudes in the van were like-

    12. JH

      Just turn it off.

    13. EH

      ... "Sounds like they're going to sweep the office for bugs. Let's turn off our giant microwave antenna." And they kept at it for seven years.

    14. JR

      Wow.

    15. JH

      It was only ever discovered because there was this, like, British, ah, radio operator who was just, you know, doing his thing, changing his dial, and he's like, "Oh, shit. Like, is that the ambassador fucking talking?"

    16. EH

      Just randomly. So- so the thing is... Oh, and actually, sorry, one other thing about that. If you heard that story and you're kind of thinking to yourself, "Hang on a second," um, they were shooting, like, microwaves at our ambassador 24/7 for seven years.

    17. JR

      Whoa.

    18. EH

      Doesn't that seem like it might, like, fry-

    19. JR

      Give you cancer?

    20. EH

      ... his genitals or something? Yeah, or something like that.

    21. JR

      You're supposed to have a lead vest. (laughs)

    22. EH

      And the answer is-

    23. JH

      A jock.

    24. EH

      ... yes.

    25. JR

      Yes.

    26. EH

      Yes. And this is something that came up in our investigation just from every single person who was like, who was filling us in and who- who dialed in and knows what's up. They're like, "Look, so you got to understand, like, our adversaries, if- if they need to, like, give you cancer in order to rip your shit-"

    27. JH

      They'll fucking give you cancer, yeah.

    28. EH

      "... off of your laptop, they're going to give you some cancer."

    29. JR

      Did he get cancer?

    30. EH

      Uh, I don't know specifically about the ambassador-

  8. 37:55 – 46:40

    Information warfare at scale: bots, audience capture, and persuasion as a threat vector

    1. JH

      But now there's also, there's so much more tooling too, right? If you're a nation state and you want to fuck with people and inject narratives-

    2. JR

      Mm-hmm. Oh, yes.

    3. JH

      ... into the ecosystem, right? Like, the whole idea of autonomous AI agents too, like having these basically, like, Twitter bots or- or whatever bots. Like, a lot of... One thing we've been, we've been thinking about too k- on the side is, like, the idea of, um, uh, you know, audience capture, right? You have, like- like, big people with high profiles and kind of gradually steering them-

    4. JR

      Yes.

    5. JH

      ... towards a position by creating bots that, like, through comments, through upvotes-

    6. JR

      Right. 100%.

    7. JH

      You know, it's like... What are the-

    8. JR

      It's absolutely real.

    9. EH

      Yeah. And a couple of the- the big, like, a couple of big accounts on X, like, uh, that- that we're in touch with have sort of said like, yeah, especially in the last two years, it's actually become hard, like, especially the thou- thoughtful ones, right? It's become hard to, like, stay sane, not- not on X, but, like, across social media, on- on all the platforms. And that is around when, you know, it became possible to have AIs that can speak like people, you know, 90%, 95% of the time. And so you've, you have to imagine that, yeah, adversaries are- are using this and doing this and pushing the frontier.

    10. JR

      No doubt.

    11. EH

      Like, there's-

    12. JR

      They'd be fools if they didn't do it.

    13. JH

      Oh, yeah, 100%. 100%.

    14. JR

      You have to do it because for sure we're doing that.

    15. JH

      And th- this is one of the things where, um, you know, like, it used to be... So, OpenAI actually used to do this assessment of their, uh, AI models as part of their- their kind of what they call their preparedness framework, that would look at the persuasion capabilities of their models as one kind of threat vector. They pulled that out recently, which they've, it's kind of like...

    16. JR

      Why?

    17. JH

      You can argue that it makes sense. I- I actually think it- it's, um, it's somewhat concerning because one of the things you might worry about is if these systems, sometimes they get trained through what's called reinforcement learning. Potentially, you could imagine training these to be super persuasive by having them interact with real people and convince them, practice at convincing them to do specific things. Um, if that, if you get to that point, you know, these- these labs ultimately will have the ability to deploy agents at scale that can just persuade a lot of people to do whatever they want, including pushing legislative agendas, vote, like, electoral...

    18. EH

      And even help them, help them prep for meetings with, uh, the Hill, the administration, whatever. And like, "How should I, like, convince this person to do that?"

    19. JR

      Right.

    20. EH

      Like...

    21. JH

      Yeah.

    22. JR

      Yeah. Well, they'll do that with text messages, make it more businesslike.

    23. EH

      Yep.

    24. JH

      Yeah.

    25. JR

      Make it friendlier, make it more jovial. But you-

    26. JH

      But this is like the same optimization pressure that keeps you on TikTok, that same, like, addiction. Imagine that applied to, like, persuading you of some, like, some fact, right?

    27. JR

      Yeah.

    28. JH

      That's like a...

    29. EH

      On the other hand, maybe a few months from now, we're all just going to be very, very convinced that it was all fine.

    30. JH

      (laughs)

  9. 46:40 – 54:30

    Securing a ‘Manhattan Project’ for AI: stovepipes, ego, and integrating elite security expertise

    1. JR

      Yeah.

    2. JH

      ... that's part of the trade-off. And that's actually a big, big part of the challenge, too. So when, when we're working on this, right? Like, one of the things Ed was talking about, these, like, th- 30 different layers of security access or whatever. One of the consequences is you bump into a team at ... So, so the, like, the teams we ended up working with on this project were folks that we bumped into after the end of our, our last investigation, who kind of were like, "Oh, uh ..."

    3. EH

      We talked about last year, yeah.

    4. JH

      Yeah, yeah, yeah. Like, looking at AGI, looking at the national security, um, kind of landscape around that. And a lot of them were, like, really well-placed. So it was like, uh, you know, Special Forces guys from tier one units, so, like, a SEAL Team Six type thing. And because they're so ... Like, in that ecosystem, you, you'll see people who are, like, ridiculously specialized and competent. Like, the best people in the world at doing whatever the thing is, like, to, to break the security. And they don't know often about, like, another group of guys who have a completely different, uh, capability set. And so what you find is, like, you're, you're indexing, like, hard on this vulnerability, and then suddenly someone says, "Oh, yeah. But by the way, I can just hop that fence." Like, it's ...

    5. JR

      Oh, wow.

    6. EH

      So the really funny thi- ... The really funny thing about this is, like, most or even, like, almost all of the really, really, like, elite security people kinda think that, like, all the other security people are dumbasses, even when they're not.

    7. JH

      Or, or, like, yeah, they're, they're, they're, they're biased in the direction of ... 'Cause it's so easy when everything's s- ... like, stovepiped.

    8. EH

      But, but so most-

    9. JH

      Yeah.

    10. EH

      ... people who, who say they're, like, elite at security actually are dumbasses. 'Cause, 'cause s- ... Most security is, like, about checking boxes and, like, SOC 2 compliance and, and shit like that.

    11. JH

      But, yeah. It, it's-

    12. EH

      But-

    13. JH

      What it is, is it's like, so everything's so stovepiped-

    14. EH

      Yeah.

    15. JH

      ... that, that you don't ... You literally can't know what the exquisite state of the art is in another domain. So it's a lot easier for somebody to come up and be like, "Oh, yeah. Like, I'm actually really good at this other thing that you don't know." And so figuring out who actually is the ... Like, we had this experience over and over, where, like, you know, you run into a team, and then you run into another team. They have an interaction. You're kinda like, "Oh, interesting." So, like, you know, th- like, these are, are the really kind of, um, the people at the top of their game. And that's been this l- ... a very long process to figure out, like, okay, what does it take to actually secure our critical infrastructure against, like, CCP, for example, like, Chinese attacks, if we're, if we're building a superintelligence project. And it's, it's this weird, like, kind of, uh, challenge because of the stovepiping. No one has the full picture, and, like, we don't think that we have it even now. But definitely don't know of anyone who's come, like, that ... Like, this close to it, at least.

    16. EH

      Yeah. The best people are the ones who when they, when they encounter another team and, and other ideas and start to engage with it, are like ... Instead of being like, "Oh," like, "You don't know what you're talking about," who just, like, actually lock on and go, like, "That's fucking interesting. Tell me more about that."

    17. JR

      Right.

    18. JH

      Yeah.

    19. JR

      People that have control of their ego.

    20. JH

      Yes.

    21. EH

      Oh, 100%.

    22. JR

      With everything.

    23. EH

      The best of the best-

    24. JR

      Everything in life.

    25. EH

      The best of the best, like, got there by ...... ab- eliminating their ego as much as they could.

    26. JR

      Yeah.

    27. JH

      Yeah.

    28. EH

      Always the way it is.

    29. JH

      Yeah. And it's, it's also like, the, the fact of, you know, the 30 layers of the stack or whatever it is, of all these security issues, means that no one can have the complete picture at any one time. And the stack is changing all the time. People are inventing new shit, people, things are, are falling in and out of... Um, and- and so, you know, figuring out what is that team that can actually get you that complete picture is an exercise, A, you can't really do. It's hard to do it from the government side, because you got to engage with data center building companies. You got to engage with the AI labs, and- and in particular, with like insiders at the labs who will tell you things that, by the way, the lab leadership will tell you the opposite of in some cases. And so, like it- it's just this, this Gordian knot, like, where it like, it took us months to, to like pin down every kind of dimension that we think we've pinned down at this point.

    30. EH

      I'll give an example, actually, of- of that, like, the- the trying to do the handshake, right, between different sets of people. So, we were talking to one person who's, um, who's thinking hard about data center security, working with like Frontier Labs on this shit. Um, very much like at the, at the top of her game, but she's kind of from like the academic space, kind of Berkeley, like the av- avocado toast kind of side of the spectrum, you know?

  10. 54:30 – 1:04:46

    Deterrence, escalation, and “going on offense”: why perfect defense is a fantasy

    1. EH

      And that, and that means you actually have to go on offense from the beginning. As... 'Cause like the truth is, and this came up over and over again, there's no world where you're ever gonna build the perfect, exquisite fortress around all your shit, and hide behind your walls like this forever. That just doesn't work, because no matter how perfect your system is, and how many angles you've covered, like your- your adversary is super smart, is super dedicated. If you see the field to them, they're right up in your face, and they're reaching out and touching you, and they're trying to see like what- what your seams are, where they break. And that just means you have to reach out and touch them from the beginning. 'Cause until you've actually like reached out and used a capability and proved like, we can take down that infrastructure, we can like disrupt that- that cyber operation, we can do this, we can do that, you don't know if that capability is real or not. Like you might just be like lying to yourself...

    2. JR

      Right.

    3. EH

      Mm-hmm.

    4. JH

      And like, "I can do this thing whenever I want," but actually... You're kind of more in academia mode than like startup mode-

    5. EH

      Exactly.

    6. JH

      ... 'cause you're not making contact every day...

    7. EH

      Exactly.

    8. JH

      ... with the thing, right? You have to, you have to touch the thing. And th- there's like, there's a related issue here, which is, um, a kind of like willingness that came up over and over again. Like one of the-

    9. EH

      Yes.

    10. JH

      ... kind of gurus of this space was like... Or m- made the point, a couple of them made the point that-Um, y- you know, you can have the most exquisite capability in the world, but if you, if you don't actually have the willingness to use it, you might as well not have that capability. And-

    11. EH

      Hmm.

    12. JH

      ... the- the challenge is, right now, w- w- China, Russia, like our adversaries pull all kinds of stunts on us and get consequences.

    13. EH

      Particularly during the previous administration. This was a huge, huge problem during the previous administration, where you actually, you actually had, um, sabotage operations being done on American soil by our adversaries, where, uh, you had administration officials, as soon as like a thing happened, so there were, for example, there was like, um, f- four different states had their 911 systems go down, like, at the same time, different systems, like unrelated stuff, but it was like, it's- it's this stuff where it's like, "Let me see if I can do that. Let me see if I can do it. Let me see what the reaction is. Let me see what the- the chatter is that comes back after I do that." And one of the things that- that was actually pretty disturbing, uh, about that, was under that- un- under that- that- that, uh, administration or regime or- or whatever, the response you got from the government, right out the gate was, "Oh, it's an accident." And that's actually unusual. The proper procedure, the normal procedure in this case, is to say, "We can't comment on an ongoing investigation," which we've all heard, right? Like, the- "You can't comment on blah blah blah." Right. We can neither confirm nor deny. Exactly. It's all- all that stuff, and that's- and that's what they say typically out the gate when they're investigating stuff, but instead coming out and saying, "Oh, it's just an accident," is a break with procedure. Um, they- i- if- if they say- if they- if they leave an opening or say, "Actually, this is an adversary action. We think it's an adversary action," they have to respond. The public demands a response. And they don't- they're- they were too-

    14. JH

      There's a fear of escalation.

    15. EH

      ... f- fearful of escalating.

    16. JH

      So- so what ends up happening, right, is- and- and- and by the way, that- that thing about like, it's an accident, comes out often before there would have been time for investigators to physically fly on site and take a look. Like, there's no logical way-

    17. EH

      Mm-hmm.

    18. JH

      ... that you could even know that at the time, and they're like, "Boom, that's an accident. Don't worry about it."

    19. EH

      So they have an official answer, and then their response is to just bury their head in the sand and not investigate?

    20. JH

      Right. Because if you were to investigate, if you were to say, "Okay, we looked into this. It actually looks like it's fucking, like, country X that just did this thing-"

    21. EH

      Right.

    22. JH

      ... if that's the conclusion, it's hard to imagine the American people not being like, "Um, what are we- like we're letting these people, uh, injure our American citizens on US soil, take out, like, US, uh, national security, like, or- or critical infrastructure, and we're not doing anything?" Like, the concern is about this, like, w- we're getting in our own way of- of thinking like, "Oh, well, escalation is going to happen," and boom, we run straight to, like, "There's gonna be a nuclear war. Everybody's gonna die." It's like, w- when you do that, you're- peace between nations, stability does not come from the absence of activity. It comes from consequence. It comes from-

    23. EH

      Exactly.

    24. JH

      ... just like if you have, you know, a- an individual who misbehaves in society, there's a consequence and people know it's coming. You need to train your counterparts in the international community, your ad- your adversary, um, t- to not fuck with your stuff because-

    25. EH

      Can I just- can I stop you for a second? When- when- so are you essentially saying that if you have incredible capabilities of disrupting grids and power systems and infrastructure, you wouldn't necessarily do it, but you might try it to make sure it works a little bit? Exactly.

    26. JH

      Yeah.

    27. EH

      And that this is probably the hints of some of this stuff 'cause you kinda do- You gotta get your reps in, right? Right.

    28. JH

      Yes.

    29. EH

      You gotta get your reps in. Yeah.

    30. JH

      It- it's like- it's like, okay, so suppose that like- that I went to you and was like, "Hey, I- I bet I can kick your ass. Like, I- I bet I can like friggin' slap a rubber guard on you and like do whatever the fuck," right? Um, a- and you're like-

  11. 1:04:46 – 1:09:08

    China inside the system: coercion of diaspora, insider risk, and clearance blind spots

    1. JR

      Right.

    2. JH

      ... basically just sitting ... Right now, our st- the state of security is ... The labs are, like, super p- And we're like, we, we can and probably should go i- i- deep on that piece, but, like, as one data point, right? So there's, like, double digit percentages of the world's top AI labs, or America's top AI labs, um-

    3. EH

      Of employees.

    4. JH

      ... of employ- uh, employees that are, like, Chinese nationals or have ties to the Chinese mainland, right?

    5. JR

      (laughs)

    6. JH

      So that's, that's great. Why don't we build a Manhattan Project with-

    7. EH

      Yeah, it's really funny, right? (laughs)

    8. JH

      Like, so, so this-

    9. JR

      (laughs) That's so stupid.

    10. JH

      But it's, it's also like, it, it's, uh, the, the challenge is when you talk to people who actually ... Uh, geez. When you talk to people who actually have experience dealing with, like, CCP activity in this space, right? Like, there's one story that, that we heard that, uh, is probably worth, like, relaying here is, like, this guy, uh, from, uh, from an intelligence agency was saying like, "Hey, so there was this power outage out in Berkeley, California back in, like, uh, 2019 or something, and the internet goes out across the, the whole campus. And so there's this dorm and, like, all of the Chinese students are freaking out because they have an obligation to do a time-based check-in and basically report back on everything they've seen and heard to basically a CCP handler type thing."

    11. JR

      Right.

    12. JH

      "And if they don't, like, hmm, maybe your mother's insulin doesn't show up? Maybe your, like, brother's travel plans get denied? Maybe the, uh, family business gets shut down?" Like, there's the range of options that this massive CCP state coercion machine has. This is like, they, you know, they've got intern- like, software for this, like, this is a- an institutionalized, like, very well developed and efficient framework for just ratcheting up pressure on individuals overseas, and they believe the Chinese diaspora overseas belongs to them. If you look at, like, what the Chinese Communist Party writes in its, like, in its written, like, public communications, they see, like, Chinese ethnicity as being a green ... Like, is it, like, no one is a bigger victim of this than the Chinese people themselves who are abroad, who, uh, made amazing contributions to American AI innovation. You just have to look at the names on the frigging papers. It's like, these guys are, are wicked. But the problem is, we also have to look head on at this reality. Like, you can't just be like, "Oh, I'm not gonna say it because it makes me feel funny inside." Someone has to stand up and point out the obvious that if you're going to build a fucking Manhattan Project for super intelligence, and the idea is to, like, be doing that when China is a key rival nation state actor, yeah, y- you're gonna have to find a way to account for the personnel security side. Like, at some point someone's gonna have to do something about that.

    13. EH

      And it's like, you can see they're, they're, they're hitting us right where we're weak, right? Like, America is the place where you come and you remake yourself, like, send us your tired and your, your hungry-

    14. JH

      Yeah.

    15. EH

      ... and your poor. And-

    16. JR

      Which is true and important.

    17. EH

      It's true and important-

    18. JR

      Yeah.

    19. EH

      ... but they're playing right off of that because they know that we don't, we just don't want to look at that problem.

    20. JR

      Yeah. And Chinese nationals working on these things is just bananas. The fact they have to check in with the CCP.

    21. JH

      Yeah.

    22. JR

      And, and are they being monitored? I mean, it, how much can you monitor them? Or how ... What do you know that they have? What, what equipment have they been given?

    23. EH

      Oh. You see, and constitutionally, right?

    24. JH

      So the best part... Yeah, the best part-

    25. EH

      You can't just spy on-

    26. JH

      ... constitutionally, you, you ca- you ... It's also you can't legally deny someone employment on that basis.

Episode duration: 2:47:55

Install uListen for AI-powered chat & search across the full episode — Get Full Transcript

Transcript of episode Wwp1BFEw3cA

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

iOSAndroidClaudeChrome