Balaji Srinivasan: How AI Will Change Politics, War, and Money

25:15We're in the age of the phrase, which is the prompt, the 140 character tweet, and the 12 words for your crypto password, right? These phrases of power in AI, in social media and in crypto, just unlock everything. So the better your vocabulary, the more you can do, right? So I think of prompts as tiny programs and actually one of the things that I've gotten in the habit of doing is writing, it's total opposite of search. You know, with search, you learn to type things in key wordes and you sort of figure out the word that has the most specifistic T F IDF, you know, on the page or whatever. I will write sometimes these long memos to an AI and then continuing, maybe you don't love this analogy, but I think it's funny.

25:56continuing the policy of scalyche I'll give them to broma vishnu and shiva okay so I give them to chat chats you to and clawed and now grok and what have you right I'll consult all the gods and then I'll make my decision on that basis right and then sometimes have them argue with each other right and why do I see kind of half jokingly gods because like the Hindu kind of frame on that is not like like the, you know, fearing God things, not the same kind of thing. But in a sense, it is a superhuman intelligence that knows everything about your culture. And if you ask it the right question, it can tell you something that you didn't know.

26:36But in like the institution, they're not infallible, right? Like, let us just say, it's not the same as the all knowing all seeing. It's like more like superhuman, some more like superiors. People will argue with me on about that, but I think that's more true. Yeah, I think that's good. It's like actually the North tradition is similar to the Hindu tradition in some ways, right? Where the gods were not involved with their superhuman. This is a great and it's a useful framing. Just remember that when it comes to computer systems, we can put formal bounds on them. We can do this information theoretically.

27:02We can do this computationally. Totally. And that's going to come. And once that happens, we will understand these systems fully. And it'll be very hard to think of them as gods at that point. Well, I mean, that's the thing is actually what's interesting is that the interpretability work that in throw up pick and others have done, and the work on like rocking or would have you, right? That's actually really good stuff because you can pick apart neurons, you can find the golden gate neuron if you saw that kind of thing, right? You can dial that up, dial that down. You can start actually taking apart these AI brains in a way that hasn't happened before.

27:33Few other kinds of things. So the thread that you guys, that thread actually summarized several of my research. I should probably put this into a post so it's like kind of there for the record. So I'm just going to do a bunch of these that maybe get your thoughts. So first concept in no particular order, AI doesn't do it end to end, it doesn't middle to middle. So the business spend, so basically you have to still prompt it, and then you have to verify it. And people talk about prompting, but they talk less about verifying. And Carpati and I had this good conversation a few weeks ago where basically AI is going to create massive numbers of jobs in proctoring and verification.

28:12Because it's so good at faking things. So one of my other kind of concepts is AI makes everything fake and crypto makes it real again. Because AI has a probabilistic technology and cryptos is a deterministic technology. And so like crypto is in some sense what AI can't fake. It's like the hard cryptographic equations. It can't fake a Bitcoin private key. It can't fake even an on -chain NFT. That's what AI cannot fake. And so that's like the hard barriers, right? I mean, generally, I mean, I don't think crypto solves the grounding problem, right? I mean, it's a mechanism you could use, but it's a mechanism.

28:48Grounding in reality? Yeah, yeah. Yeah. Okay. The data ingest problem. Yeah. So I just figured you know that here's why there let me give you a counter argument at least. Right now, just give a concrete example. Let's say you asked perplexity to summarize the FTX hack in 2022. when it would do so among the citations that would give you would be links to a block explorer, right? That would actually have on -chain data that you can cryptographically verify that this transfer of these funds happened at this time and if you want to go even further, you can actually pull out the digital signatures and the hashes and the time stamps from that block explorer, right?

29:26Okay. Now, here's my argument. My argument is that works for financial data, But what's happening now with Farcaster and other kinds of things is with the increase in block space, you could put more and more kinds of data on chain and we're going to have to because you're going to need crypto instruments and you're going to need cryptographically hashed posts and crypto IDs to know that it was posted by a human or to know the data wasn't tampered with. So more and more kinds of data are going to go on chain and then that will eventually mean that an AI's citations are to on -chain data, which is both financial data and social data.

30:02And so then at least it will map back in terms of grounding to an on -chain cryptographically -prooable assertion of some kind. And you might say, well at least that'll be an assertion at the metadata level. Like we can prove that this digital signature made this assertion at this time stamp with this probability. Sure. I'm talking about real world grounding. Like, I say something, I am a human being, you know, you have no idea what I said is true or not true. You know, there's a geographic place where there's a picture of the geographic place taken from a 1970s photo, was that doctor or not? I mean, like, like the actual physical world grounding, just because you can't encode, you know, digital data yet at this point for the physical world.

30:45Now it's a great mechanism to do that once we can solve the ingest problem, but this just... So let me talk about something which is happening now that I've been kind of funding on the side. It's not a full solution, but it's a, I think, a partial solution, which is... So crypto instruments, the idea would be that when you capture like a frame of data, right, for example sequencing machines, like DNA sequencing, when the data is coming off the machine, it's like TIF files that are actually image data that gets processed into AC, Gs, and T's. And many other kinds of instruments basically have a stream of data coming off the machine as you're capturing it.

31:21Cameras are like that, right? So you could, and there are things to do this already, take a hash of that and post it on chain at that time, right? And what that would at least say is that that frame of data existed at that time. And so if you had something that was like a scientific experiment, right? You know, like a pre -registered double -blind trial or something like that. You could have, not just a cryptid instrument, you could also have other people with proof of humans there who have a sort of attestation ceremony. And now you have a number of different kinds of on -chain data that start to get harder to fake and coordinate.

32:00Wait, not impossible. Totally. I agree with all of it. As soon as you get it into the system, then crypto is a great mechanism. and for ensuring kind of end -to -end guarantees. It's just the data -in -just problem is, it's a long -standing problem. That's right, computer science. And over time, everything you're saying is gonna be more and more true, because over time we're gonna be more and more incentive to make sure the stop going into the system is true. And so I just totally disagree. It's just, listen, I'm an old -school networking guy. Like, for us, there's like the internet and there's the stuff that goes in the internet and then you just kind of use different mechanisms for both.

32:30And this is what's calling out. That's all. Okay, great. So next kind of concept maybe to discuss. And I think this is a useful division. This is a relatively recent point that I made to myself that that was useful. AI is good for the visual and less good for the verbal. What do I mean by that? So when it's generating images, when sharing video, when sharing user interfaces like Vercels V0 or a Replicit user interfaces, the great thing about them is you can instantly see them. And with the GPUs that we have in hardware, you can verify cheaply whether they're good enough, right? Because you can just instantly get the gestalt of it, right?

33:08Whereas when it's back end code, when it's legalese, when it's like, you know, mathematical equations, you have to slow down and use system two thinking, not system one, right? And it's not just your gestalt impression. You have to actually go line by line and check whether it's right. And that is actually the expensive step. the verifying right over there. So I think that's a non -obvious thing where the more front end and video and visual it is what you're doing, the easier it is to say, and now the interesting concept is how much do I go ahead and say? I'm going to say that on one other thing, which is like for me again, like I spend most of my time in software and engineering, the big distinction of stainless versus stateful.

33:50Right, so if you're generating code that it's going to have some and the same magic that evolved while you're running it, it's just impossible to spot check. Like some things are computationally reducible, you actually have to run the computation to the other answers. Like the image of the perfect example, it's visual, and it's basically stateless, like all of the state is there. There is no kind of runtime cement, totally agree. That's right. And whereas even a relatively small snippet of backend code could have a fairly complex finite state machine essentially underlying it, or even infotainment state machine.

34:20But you can't just dynamically bow. Yeah, it can be very, that's right. And so, simulating the tie and dynamics that you have to see something different, maybe formal verification, if it's algebra, you know, it's all, yeah. Where you actually have to run it if it's computation irreducible, like there's no way to statically do it. And so it does reduce to almost like the computation verification problem, which is this longstanding problem in computer science forever. Right. Now, formal verification, at least for a subset of programs, has become commercially viable for smart contracts because they're so high value and they're so small that they actually, it's worth doing that on, right?

34:53Totally. And yeah, it's not gonna work for the general case, but you can do a constraint case. But, okay. So that was one major division, visual versus verbal. Another, when you're getting to like, stateful and so on, is the limits of AI are the things I'm interested in where you draw like a fine distinction on what you can do, are very crisp. And what I think AI is particularly bad at the people that are trying to use it for, that they're gonna fail on in my view, is when they try to use it for markets or politics, And let me explain why I say that. For systems that are time invariant, you know, like the mapping of an image to the label cat or the rules of a game like chess or checkers or even go, or you know something where there's like a static rule set or static mapping, right?

35:41Then you can do the train test paradigm and train a model and so it's over. However, when you have something which is time varying, especially rule varying, an adversarial like markets are or like politics are, then the same trade will eventually quickly start resulting in a loss. And by the way, the other guys are also using an AI on you, right? And so it's decentralized AI again, right? And so that does actually argues that the CEO or the creator who is constantly sensing the market, resensing the political wins, and as a thesis on it, based on human nature or other things, or what have you, actually is the sensor that then prompts the AI, and that's a job that is hard for the AI to do at a really deep level, because it's time varying, rule varying adversarial demands.

36:33And it goes back to what you said in the very beginning, which is if you look at these type of equilibrium, there are complex differential equations which are nonlinear.

36:43Yeah. extrapolation, which we know that these things are not very good at. What's interesting is I wasn't even thinking of the stock market as complex different, but you're right. But it's a chaotic system. The chaotic system. I mean, you wrote this great book on the fact that these things are super chaotic. Yeah, you're actually right that actually it would be a useful thing to show just with a toy example, a chaotic system that is time -aving or another one's out of Cheryl's adeniac. Now, the thing about that though is to argue against my point, they've got an AI that are actually pretty good at starcraft, right?

37:17Which it starts to stretch the boundaries of what I was saying because it's definitely adversarial, right? And it's like more time varying than chess, you know, you could argue it's not rule varying, but it's time varying. Okay, let me go to another point here, right? So another concept is I think, so maybe the commercial implication of that point on prompting and verifying is business spend moves towards prompting, proctoring, verifying, basically checking all the stuff that AI can generate. That's going to be a huge, huge, huge thing. And that maps to KYC, that maps to like in a bad way, you know, the glass cases in Walmart right?

37:57In a sense, a lot of society is spending more and more and more on verification and proctoring and so on, right? Okay. Next, AI means amplified intelligence, not agentic intelligence, because the smarter you are, the smarter the AI is better writers or better prompters. What are your thoughts on that? Yeah, I mean, it's interesting in the coding space that we actually start to have numbers on this now. Oh, interesting. Yeah, so if you actually look at relative productivity gains, it just turns out of you're a more senior developer, you will have better productivity gains and the AI. I haven't seen that graph.

38:29So it is something that makes the smart smarter, basically. But also on a relative basis, which is really surprising, right? If you think about it, it's actually not surprising. It's like, you know what the fundamentals for you to ask, you know how to interpret the results, you know how to throw away bad stuff when it's bad. And so clearly, if you kind of know what you're doing, you can both verify to your point the output, but you can also be more specific of your ass. I think it's really important for all of us to realize that formal language just came out of natural language is not the other way, right?

38:59Like if you could explain all of this stuff in English to each other, we would, but it's just really inefficient. So we came up with more efficient ways to train. Yeah. The more trained languages that reduce ambiguity. This literally is strictly an efficiency thing, right? And so like some of that knows how to speak these formal languages to the models is going to articulate, but they want better. And this can be able to interpret the results better if the response is formal. And so it's kind of a nice codification of exactly what you're saying. Yeah. I mean, the thing about it is AI means everyone's a CEO.

39:29because to a great AI, you speak to them in some ways to a great employee where you give clear written instructions and then you can verify the output. So it actually turns management into a skill that it hyper deflates the cost of trying one's hand as a CEO or as a manager because you have to give those. And so the better you are in communicating what it should do, often the more people you can manage and so on and so forth. But this gets sick the next point, which is AI doesn't really take your job. It takes the job of the previous AI. Okay. And what I mean by that is you now have a slot on your roster at every company for an AI image editor, an AI text, you know, chatbot thing, an AI code, you know, IDE thing and so on and so forth.

40:17And each new release of GROC or Clawed or whatever, competes against chat, CBT and GROC and the cloud, right? And so the AI takes a job at the previous AI, because they're complementing, you kind of have a whole raft of AI augmenters that are augmenting your humans. But those AI's are competing in AI space to a large extent with the previous AI, because once you've onboarded an image generator into your flow, then it just keeps improving, and you start using it in more places, but it's an AI taking the job of the previous AI. Let me know your thoughts. This is an adjacency to what you're just saying, but can I actually push on something you said previously?

40:56So I actually agree with your polytheistic view of the world. I totally agree, but let me just provide the counter argument for us to do the lawn in this vein, which is have you seen this kind of thing that all the AIs that you asked to produce a random number produce the same number? Have you seen this? Yes, it's like seven or seven. Seven or seven. Right. So one thing that is to me was not intuitive, but remarkable about these models is how easy they are to distill, which is as soon as it creates a leader, everybody uses that leader and kind of sucks the life out of it. And then all the models turn version at very quickly, which you could argue that this is a counter to the colon - Yeah, colon - basically D.

41:35Yeah, there's like a core. Maybe there are 100 AIs, but it just turns out they all have the same capability. So is it just a technicality that they're actually different than they've all learned from each other? Well, so this is an interesting question. And my view is, I'm not called a strong view yet, But my view is that's almost like the human body plan and spinal column and then you differentiate on top of that Core spinal column maybe, you know, and it's like you'd have some sort of just like every human to first order can See and speak and hear and so it's up for it, but then some people have You know much better vision or they have much better speech or something like that, right?

42:14So there may be some distilled kind of thing. Oh by the way another interesting part on what you're saying. In general, text on the internet is not emitted equally by every group. I mean, here's how I would distill my view on this, which is very much in line with what you're saying, which is, I think the universe is very complex. And I don't think it gives up its secrets easily at all. And I think the universe is full of fundamental trade -offs. Like, you can't have both. You have to choose A or B. And so these models will align with those fundamental trade -offs, right? And maybe it's performance, Maybe it's correctness, maybe whatever it ends up being.

42:53And so as soon as you want a specific solution for a given problem, where it's one of those trade -offs, you're just gonna need a different model because you just can't end up having both. And again, because I know the coding space the best, we see this a lot, right? Which is a model that's very good for certain parts of code is just not gonna be generally good at other things because those are the trade -offs made when training it. And I think that this is the future plurality of model. We're going to. Well, is that true? I, you know, I thought somebody said something. I may be wrong with this, but I saw some countertune result that said that making the AI specialized in one area makes it worse in other areas.

43:29Do you see something like that? Well, yeah. So this is a very big debate. But the debate goes as follows. Like the first wave of AI was pre -training where everything you threw into it, like it just got smarter. And so that's kind of a 10 for 10 technical win, right? Just because it'll be as good as writing code as it saw it. But as soon as you're doing RL, where you're training it in a specific domain or the specific garifier, you're likely losing other areas. So you make it really good at playing chess. It's going to be less good at something else like that, writing general scores or something.

44:05So I think the current debate now and the current data seems to suggest that RL doesn't generalize in the same way. And so now we are in this case where you would have a plurality of models because you're always robbing Peter to pay all when you make it good at a certain domain. That's right. Yeah. I think also, you know, someone relates to that in terms of what demands it's good at and so on and so forth. At least right now, I think I'm sure if you agree with this AI doesn't really take your job, it allows you to do any job. Yeah. Because you can get to like an OK level as like a user interface designer or sound effects or something like that, but you need a specialist for polish and that, you know, though I wonder maybe with enough RLHF from specialists, maybe that won't be as necessary.

44:52I don't know, maybe you have some thoughts. So here's my current mental model. There's two personas. This persona number one is the expert in the space and this persona number two is the non -expert in the space. So if the non -expert in the space is using AI, it's taking the place of the expert, right? And so maybe you'll ask it and it'll give you some. Listen, I want a 3D asset for a video game and I'm a programmer, then I'm going to ask it for a nice 3D asset and then it'll give me one, right? So I'm the non -expert. The expert user to our previous will actually know how to ask it better and will likely get better results because it's an actually domain expert.

45:29And that expert is using it. So I think we see both of, actually, you look in the market, you see both of these uses. And I think both of them will persist, which is if I'm a programmer, it's like a doctor talking to their doctor and they can just instantly go to a specialist language and so forth. So you're right. And even if they're specialist RLHF, you may not be able to access that. Yeah, that's right. Why would I want to learn the entire domain and make all of the trade -offs of 3D design that somebody else could have done all of that work for me, and they could talk to the model in this specialist way when I can just talk to my model using code.

46:06Right? In order to very efficiently use a specialist model, I would have to become a specialist, would be the argument to our previous one. So I think for casual use, I can use these models for whatever I want, but to really use them very well again to our previous conversation, I'd have to become a specialist and somebody else will have maybe already invested all of that time. By the way, if you actually look at products, these products actually have both these distinctions. Some products are very clearly for the casual user, trying to replace like, you know, think about cursor versus lovable, right?

46:36So lovable is I'm a casual user. I'm going to create a website. I don't need to have to know about code. And that's great, great amazing things. Cursor is I am a professional software developer. I have an IDE and an IDE. Now, over time, maybe these things converge. That could be the case. but thus far, these are very different user bases, right? There's professional coding versus basically casual coding. I mean, you know, part of it is, which is interesting, and a little counterintuitive, it's like, if you think about what a computer can do, it can do the job of counting, it can do the job of physicist, but then you clad it in something and it's adding up numbers in Excel, and you clad in something else and it's doing simulations from MATLAB, right?

47:18And so we already know that when it came to logical system 2 thinking that computers are actually really good at that. And now you have something similar where these models are actually very versatile, but you clad it in the power user interface and you clad it in the casual interface. And it's implicit contextual prompting as well, probably as well as the system prompt that makes it do those things. I mean, the thing that's interesting to me about something like chain of thought is that And I think this is where people were freaking out in late 22 and maybe they'll still be right to freak out is Computers have historically always been good at the logical style much better superhuman at that Now they're also Superhuman in a sense at the probabilistic style at least of text generation and so and so forth And so it's not inconceivable that someone could figure out a way to merge those two You know like a quantum gravity theory of things right where you take the probabilistic and deterministic and pull them together.

48:18Yeah, this is where the very old school, you know, systems part of me thinks that there's a fundamental trade -off here, right? You can trade off. Well, I feel like you can trade off. You can build a system for determinism. And you can, you know, build a system which basically cuts a bunch of corners. But you can't build a system that does both. I mean, I can give you stuff. What I mean, it has gotten pretty good at figuring out when it wants to generate an image, when it's supposed to search, when it's supposed to read a PDF and, you know, like. Right, but now you're acknowledging exactly what I'm saying, which is some things you want the fuzzy thing and some things you want a traditional system.

48:57Yes, but if you have a tool you're facing on top. Yeah, yeah, for sure, but then maybe it just becomes a consumption layer and all that hard work is still being done by traditional systems. The full argument is if you have a model that does everything, you wouldn't need tools, right? Because tools are literally like the API to the traditional system. And so that's almost like a capitulation that like some things you want, you know, traditional software to do. No, I know, but when seeing is a hybrid system, if you're just a pragmatist and you don't care, right? Could a hybrid system actually get there?

49:28I can't see it, it couldn't, right? And you know, maybe it's something where where we actually need something like Elon's, you know, billion miles of Tesla driving. If we have enough context, not from LLMs, but from pointer movements and mouse clicks on, you know, iOS or macOS. Any iOS could. Yeah, yeah. So your question is can I have one trained, well that one trained LLM that can do both like the fuzzy stuff in the hyposistate. Like is that possible? Yeah, my gut again, this is total intuition is that the universe is way too heavy -tailed. it's way too nonlinear. And so the state space is too high for that to actually encode all of that and basically it's human's can do it.

50:15No, we don't. We use calculators and we use software. Like the whole reason we're I know but at the end, we build software is because we can't do it. But at some level, maybe we're just talking about two different things. So clearly AI with traditional software can do great stuff. To me, the question is, can you have one AI that does all of those things without traditional software. Can you build one LLM and I think the answer to me is obviously no, but I've heard arguments that it can. Yeah, what's interesting to me is, so that's why I say I'm a little bit of a, I mean, I'm on the borderline of the tech pragmatists and the tech radical, right?

50:53Where I think I always want to try identify the limitations of the systems today and then see how you could push beyond it. So, I consider calculators which were made by humans to be ultimately humans doing it. You know what I mean? Like in the sense of, it's like a tool that we came up with. I mean, to give you an example like earlier, I was like, you know, AI is good at visual, but not at verbal. But for example, you can turn audio into spectrograms, which you can then look at visually, and at least you can see for radical deviations and so on. Maybe not the entire sound, but you can see. And so can we do for you transform like things on other outputs of AI where we can quickly inspect them visually.

51:36You know, is there some grid or visualization kind of thing where we can turn into a visual problem? Yeah, I love this. So what I like, I thought experiment I have is can I literally create a bunch of audio outputs so that I can listen to my AI like I listen to a car? Because like, you know what? Yeah, you can tell if it's rolling. Like, I don't know what's wrong with my car, but I know it's not normal. But very good atmospheric inputs. and I think this is a great idea as can we start exposing the internals so that we understand when it's working well and when it's not working well, you know? And I just feel like there's a lot of stuff we could push on here that, you know, we're very early getting stuff.

52:10Yeah, like colored text, for example, in terms of its level of confidence. Yeah, yeah. You know, stuff like that, like yellow red green, right? There's a lot of AIUX that one can do. Let me make a few other points that take you're interesting. Killer AI is already here and it's called drones and every country's pursuing it. So we don't have to care really about the image generators and chatbots all the worry about super persuaders or whatever is all pretty stupid Strong agree Strong agree, right and the thing is when I push people on this What's interesting is some of the people who were oh my god?

52:42We need to regulate everything are now actually on the side of we need to build it before chant but the thing is in both Senses first of a safety then a security but it both comes to control and you might argue that the security argument is a better argument I think it's a more realistic argument in some ways. But the concept of killer eye is already here, is interesting because they put so much stock in the like, oh, it's gonna persuade everybody to do things, it's a super persuader, but persuading is statistical and you know, drones are deterministic, or at least, you know, the guns on a drone or whatever a term is to go.

53:15I think that the interesting question around AI and like attack and defenses does it change the equilibrium? Because the internet did. So the internet actually introduced the notion of asymmetry, which is the more that you rely on it, the more vulnerable you are. So to it, the United States is more vulnerable than, you know, random, you know, random third world country. And it's not clear to me you get the same thing with AI. Like it could just be like, it just enables everybody to have bigger weapons, but the equilibrium is the same. Well, I think that it actually has really huge impacts for borders.

53:48And unfortunately, I think China is well positioned here for a very specific reason, which is, China, their justification for the great firewall is they've justified it as digital borders. They say we can introduce physical packets, why can't we introduce digital packets? Right? And now with the whole Ukraine controlling drones in your territory thing, that becomes more than simply a metaphor. It's a real thing. It's like controlling cloud space, right? And you know, if you can It allows somebody to script drones or script humanoids in your jurisdiction, then they can blow things up. That's no longer a theoretical thing.

54:30Now the counter -counter argument is, well, maybe you just have them pre -programmed autonomous so they don't even need an internet connection and they can just do cameras or whatever. That's true. You know these things with the drones on cable, so it's crazy things they're doing there. Right? You know those big unwinding cable things that you sometimes see on ships, right? So they have these cables for these one -way drones that are these ridiculously long, like multi -kilometer long, Ethernet cables, like Catwin cables on a drone, so it can be offline and it goes past signal jammers or something like that and then goes and blows up on its target.

55:07And it goes forward and the cable gets tangled in the trees, they don't care. because it's not going to go a reverse trip where it gets has to yank. So when one way path, it doesn't matter. So one way drawn, which is crazy. So that's an argument that at least at short distance, near the border, drones, you know, would be able to get a new one, right? But that concept of digital borders becoming hard borders, I think is going to become more of a thing where basically, I actually give a talk on this 12 years ago that your immigration policy becomes your firewall because with telepresence you can move robots around and that's starting to become real.

55:49So that is something where I think that is real implications for the geography of a country because the alternative to that of having quote -defensible borders is basically an encrypted state where you don't even know where it is on the face of the earth. And what I mean by that is, can you make a map of Bitcoin? Not really, right? It's something where it's so dispersed and you don't know every holder and they're moving around the world and there's no single map of every mine. And even if you got a map, you wouldn't know if it was complete or a RoniS or out of date or something like that. You actually have an ascent security through obscurity so you couldn't just go and blow all those things up versus something that's outlined on the map, is a recessile and vulnerable in a certain way, right?

56:36So that's something I think about a lot in terms of what do future borders look like? And you might have hard digital borders and China might preserve its territory, but those that can't enforce for rare reason hard digital borders can't stop these kind of drone, you know, interactions. Let me know your thoughts. This and some guns are not autonomous. Well, they still need to be given a control signal to do something, right? Not an entirely autonomous. You just be like, go blow up this building. Here's a picture. And then by a flake of top, we would not need any packets. That's true. That's true.

57:08And then also, if you think about it, are you going to block every single telephone call in? It's really difficult. That needle in a haystack. I have some super spooky stories of big in China. And I remember this is kind of a non -sacritor, but I have to tell you, it's so spooky. I wasn't trying to have probably 10 years. Now, I just went for the government. Nice working in the intelligence community. Like, it was a kid. I was a kid. I was a kid. It was like in 2001, you know, like my first job out of college. But anyway, so say 10 years later, I was on a business trip in China. And I called a friend of mine and I was telling about my day and the phone drop.

57:42So I called my friend again and I was telling my friend about my day and the phone drops again. And I'm like, what's going on here? So what I was recounting is where I had been. So I called my friend one more time. I said one, two, three, Tiananmen Square and the phone drops. So I think there was a bug in whatever software they had and somehow I had been picked up. But I don't think it's too crazy to assume every conversation on every phone call can be monitored and has that. Well, now they have something this is another way where AI does change balance of power in the following way, right? Like in China, they had this saying, which is the mountains are high and the emperor is far away, right?

58:24And China always had a different conception of the balance of power between the government and the people and the West. On the one hand, this is a broad generalization of a thousand years history, but very broadly, like on the one hand, because the state had all the weapons and the army and so and so forth, that chair could morph into an agent of the emperor or a CCP guy today if the government so desired, because there's no limits truly in the sense of it just do whatever whatever it wants, right? On their hand, whatever the laws are in down, the people who just do what they want, right? So they kind of, in a very pragmatic way, say, the limit is really the limit of what people can enforce, right?

59:07And at the state has lots of power than any written limit doesn't really matter. But the state can't find you, since you're on their side, the world and the written law doesn't matter either, right? Which is a different conception than the progressive versus libertarian within the West, where they'll always quote law against each other back and forth, what is written and is what is permitted or what happened, right? But AI does change that balance because now the mountains are never high and the emperors never far away. The long -larmed is incredibly long. The long -larmed is infinite. Yeah, they can synthesize.

59:39You know, there was something, maybe you know this thing, Martina, I think it was called TIA Total Information Awareness in Iraq, at a certain point. But the idea was they had satellites covering a rock and so every time some guy was putting and got in like an IED or something like that, they would rewind the satellite to find who the guy was that did that and where he came from and then put a bomb through his window or would have you, right? And in a sense, it was like tracking someone for like their whole life, you know, because you were sewing together the trace of them through all of these cameras, right?

1:00:10And that's totally possible for China to do now. And the difference is that AI makes it possible to, for a long time that data was ingested, But it couldn't really be parsed or query because it was too difficult to look through 5 ,000 hours of video on one person or whatever. That's increasingly becoming queryable, right, and ingestible and summarizable in a way that it never was. And so I think that the real check on something like that is going to have to be a cryptography exit, you know, and so and so forth, which has ultimately like get out of the jurisdiction, you know, have property that they cannot actually see.

1:00:44like you go back against the limits of power and what have you, right? Anyway, let me pause there. So some thoughts on balance of power since you talked about that. No, that's great. Oh, that's great. Okay. Last one. I think that there's going to be, there already is an anti -AI backlash that's like the anti -cryptobaclash and will be part of the anti -tech backlash because a lot of people are not using AI for what we're using it for. They're using it for like therapy, or they're using it like as a companion or something like that. You know, it's the top of the pyramid of needs. It's funny to be actually looking at like self -exualization, spirituality therapy.

1:01:21It's like finally computers are addressing like for the top of that. Exactly. And there's another aspect to it that I think it hasn't gotten as much press, but that's interesting to understand is that just like, you know, the tariffs are meant to kind of ward off Chinese competition. I'm not sure if they'll work. In fact, I'm skeptical. There's a similar, much less publicized thing that's happening at many media corporations, where they're unionizing to try to ward off AI competition. Like they have union contracts that say, the adters owners cannot use AI. They're making their organization very brittle, where they think they own the market, but they're not allowing themselves to use AI.

1:02:03And then eventually they're going to be beaten by AI -enabled competitors that pull all of their followers and views and so on away from them because they're just more efficient. So I think that's going to result in an anti -AI backlash. And I think that it's already here where, you know, on some like artist forums or whatever, they'll say, are you an AI supporter? Have you heard that? You know, like AI artists spend as much time on building things as traditional artists. It's just a different tool. This is my view. Yes, yes, yes, that's true. But basically they feel that it's similar to the reaction by master craftsmen in the 1800s, right?

1:02:41When mass production started taking over what they were doing in the physical world, this is now happening in the digital world, right? Totally. And the other aspect of this is I don't think people have thought about the international aspect where if you've got, let's say, a lawyer who's making 200K a year in the US or a doctor in the West, and then you've got somebody from the Philippines or India or anywhere in the world who's making currently $2 ,000 a year, maybe the converged wage with AI plus their IQ or an AI plus human convergence is like 20k a year, which is a 10x for the person who's abroad, but a one 10th for the person who's the last.

1:03:17And it radically increases the consumer's surplus and so forth. But I do think that that's gonna be something that's gonna be a big deal in the years to come. And so we'll have to figure out how to mitigate that. To your previous point, well, I just think this is so important. Like I agree, this can be a huge backlash. And I think some of it's gonna be rooted in the experience of individual people, my job is shifting and then I am very sympathetic. I think we should address. But I think there's something more pernicious going on, which is, and maybe this is my cynicism, but more and more, I kind of view politics as like you've got pretty sophisticated people and they have clientele classes.

1:03:49Yes, patron -pigeon client. Yes. Yeah, what they say is basically what will be the clientele class the most. That's what they just look for sound bites that will move the clientele class to actually, the patroner's sophisticated people can actually hold nuance in their head. They know that they're helping. It's dumb and down on purpose. But they dumb it down on purpose, right? And what better talking point than AI just goes back to the Promethean legend. I mean, we're terrified of technology. You can answer, borrifies it. You can talk about it just God's. I mean, it is the perfect tool to mobilize and they're seeing this on both the right and the left, right?

1:04:20So this is not in any way, to hold into one party. So I think this is like the ultimate, you know, political, you know, tool for any purpose. And we're seeing it for that. And I think for that, even more than crypto, by the way, I think the EIA strikes to the heart of people's insecurities more than crypto ever could. And so I think that this is the big battle and I think it's going to be bigger. It's interesting. It's all of the above, right? Because AI is interrupting media. Crypto is taking power over money. Robots are taking power of manufacturing and drones are taking power of the military.

1:04:51So all of these... And by the way, there's a crypto angle to at least three of them. Because I was just a crypto -angle of money. There's a crypto -angle to AI in terms of constraints. There's a crypto -angle to the drones because you're going to want the control plan for the drones to be on chain since that's a part that can't get hacked, whereas it's a pentatonic attack. So I do think this is something where it's going after quite a few power centers. I've just talked to 300 years ago, you said, well, listen, you know, we're gonna do this new crypto thing, people would be like, well, if you said listen, we're gonna create AI, these artificial, you know, intelligences that have unbound power.

1:05:21I think you're really getting out a core human insecurity that we've seen in Miss Inletian for 3000 years and probably longer. That's true. That's true. Well, we'll see what happens with currencies and so on, but I think you're right. Thanks for listening to the A16Z podcast. If you enjoyed the episode, let us know by leaving a review at ratethispodcast .com slash A16Z. We've got more great conversations coming your way. See you next time.