#173 - Liberman - We’re Training AI to Get Rid Of Us

Summary

The Peter McCormack Show – Episode #173

You’re Paying AI To Replace You
Guests: Daniil & David Liberman
Date: May 8, 2026
Host: Peter McCormack

---

Episode Overview

In this episode, Peter McCormack sits down with Daniil and David Liberman, pioneers in decentralized AI infrastructure, to discuss a critical question: Are we collectively funding and building the technological systems that will ultimately render us obsolete? Drawing parallels to Bitcoin’s journey toward decentralized money, the conversation confronts the centralization of AI power, the risks of corporate and governmental control, the erosion of jobs, and the race to create censorship-resistant, permissionless alternatives. The Libermans also share personal motivations, introducing their own decentralized AI protocol and its mission as a resistance movement against digital authoritarianism.

---

Key Discussion Points & Insights

1. Centralization of AI & Existential Risks

• AI Run by Few Corporations = Danger
  • Daria warns of a “1984-type situation” if AI infrastructure is controlled by just a few corporations or governments. ([00:00]-[02:17])
  • Lieberman describes a future where manipulation is baked into every answer:

    “We are not going to even know it’s a slavery…They own you.” ([00:11]-[02:35])

  • Peter frames the risk as an “opt-out problem”—if AI pervades everything, can anyone really live outside it? ([03:16])

2. No Escape: AI, Productivity & Dependency

• Opting Out Means Losing Out
  • Daria: “You will be able. But…whether your productivity will be on parity with people who use AI, that’s a question mark.” ([03:43])
  • AI replaces labor: even specialists have to use it to stay competitive.
• Future of Money?
  • Lieberman speculates: In a world where one AI company owns all the infrastructure, people might create parallel economies—but only if control and choice remain. ([04:18])

3. The Battle: Centralization vs. Decentralization

• Historical Parallels
  • Peter: The same struggle that shaped the US Constitution and money (Federalists vs. Anti-Federalists, central banks vs. Bitcoin) now plays out in AI.
  • Lieberman: “Not going to, it’s presenting as we speak.” ([05:57])
• Gatekeeping & Recent Events
  • Daria references OpenClaw’s shutdown as a wake-up call, highlighting “the access to AI is already right now visible that we shouldn’t take it for granted.” ([06:45])

4. AI as Addiction, Productivity Trap & FOMO

• From Tool to Master
  • Peter notes he’s “weaning off his phone” but “going deep with AI,” questioning when AI usage flips from empowerment to servitude. ([08:35])
  • Lieberman: “It’s very addictive… because you just say and stuff happens.” But this apparent freedom can breed new dependencies. ([09:34])
  • Daria: “Dependency is already much closer to the slavery.” ([12:10])
• AI Exhaustion
  • Peter: “I think I’m suffering from AI exhaustion…It’s made me more productive. I’m actually working more, more.” ([10:27]-[10:34])
  • Lieberman: “The only way to keep up with AI right now is to be unemployed, literally.” ([11:02])

5. AI, Jobs, Gig Economy & Inequality

• From Bookstores to Gig Jobs
  • Peter illustrates how tech centralization (Amazon, Uber) destroyed local ownership and upward mobility. AI could do the same.
  • Cites Eric Weinstein: LLMs “suck up all the innovation of human history and centralize the upside to the fewest group of people.” ([13:56])
  • Lieberman’s “ant” analogy: humanity as little more than moving parts for giant corporate AIs, “until the robots kick in.” ([15:31])
• Alternative Vision
  • Lieberman: The “beautiful world” is where “each of us have the robot,” with equal AI power for all, leading eventually to “superabundance”—if, and only if, access is equal. ([15:32]-[16:47])

6. Can Decentralization Win?

• Skepticism & Hope
  • Peter: “There aren’t enough people thinking about this as a problem...Centralization has won over the last 200 years.” ([16:47])
  • Daria: More people will convert when faced with layoffs and gatekeeping. Grassroots, open-source efforts like Bitcoin prove alternatives can scale. ([17:41]-[46:01])

7. Open Source, Bitcoin, & The Path to Decentralized AI

• Role Models: Linux, Bitcoin
  • Lieberman: “Everything we know about internet and…protocols today…is open source.” ([40:16])
  • Bitcoin succeeded because of both economic incentives and idealist developers.
  • Decentralized AI can work the same—if incentive structures and community are right.
• Cost Trajectories
  • Daria: “We believe that with AI will happen the same…only if we’ll have the same type of structure which constantly drives the price down.” ([45:22])

8. Decentralized AI: The Liberman Protocol

• Technical & Economic Design
  • The Libermans describe their open, token-based, decentralized protocol (Gonka) as “censorship-resistant AI.” ([64:21])
  • Key design features:
    • Proof-of-work model (not proof-of-stake; rewards infrastructure builders, not just capital holders) ([58:10]-[59:08])
    • No foundation/company; tokens distributed to miners; Libermans hold 10% (four-year vesting) ([55:09]-[57:38])
    • 20% of network power reserved for open-source model training ([49:19]-[50:18])
• Mission Over Profit
  • Peter challenges: “If you want to do this, you have to be the Satoshis of AI…mission without financial upside.” ([116:02])
  • Lieberman: Resource is crucial, but also open about the trade-offs and transparency.

9. The Race: Existential Stakes & Near-Term Danger

• Timing is Critical
  • “We maybe have one year, two years, and then it will be too late.” ([80:31])
  • Daria: Signals of centralization trend are already clear, just as Facebook ignored early warnings on Russian social media manipulation. ([108:47]-[110:07])

10. Darkest Scenarios & Final Reflections

• Authoritarian, Dystopian Future at Hand
  • Daria: “Less freedom. Like, clearly less freedom. Like everything is going to be controlled for the sake of our own safety.” ([101:19]-[101:35])
  • Peter: “You think within two, three years we’re going to descend into an authoritarian, dystopian future.” ([101:49])
• Warfare of Access and Power
  • Lieberman: “We’re not even going to get to the point of the war against the machine before we’ll start the war against the corporations which are stealing all of the jobs.” ([106:45])
  • “The darkest scenario is...there won’t be humanity anymore.” – Daria ([105:47])

---

Notable Quotes & Memorable Moments

• On Subtle Digital Control:
  • “It’s not even a slavery. We are not going to even know it’s a slavery." – Lieberman ([00:11])
• On The Addiction and Paradox of AI:
  • "In a sense you are a slave master." – Lieberman ([09:34])
  • "The only way to keep up with AI right now is to be unemployed, literally." – Lieberman ([11:02])
• On the Danger of Centralized AI Power:
  • "They own you." – Lieberman ([02:35])
• On Decentralization vs. Centralization:
  • "There aren’t enough people thinking about this as a problem... Centralization has won over the last 200 years." – Peter ([16:47])
  • “Everything we know about internet...is open source.” – Lieberman ([40:16])
• On Building an AI Alternative:
  • "When we pay for all of the AIs, we pay for the corporations to build the infrastructure and they will build it using our money to then replace us. That’s the funny irony…we keep paying them for them to replace us." – Lieberman ([90:43])
• On Existential Risk:
  • "We maybe have one year, two years and then it will be too late." – Daria ([80:31])
  • “The darkest will be the war. The actual real war…not against the machines. ...It is happening right now, that's why I'm saying like it is happening right now.” – Lieberman ([106:33]-[107:12])

---

Timestamps of Important Segments

• 00:00–04:14: Framing the problem—centralization, potential for digital slavery.
• 05:28–06:43: Historical echoes—US constitution, rise of Bitcoin, parallels in AI.
• 08:35–12:40: Personal stories of addiction, FOMO, and dependency on tech and AI.
• 13:56–16:47: Gig economy, concentration of wealth, and how LLMs centralize innovation.
• 40:16–46:01: Open-source movements, Bitcoin, and the blueprint for decentralizing AI.
• 49:19–54:23: Technical foundations of the Libermans’ protocol; design choices and tokenomics.
• 64:21–68:26: Mythos case, AI safety fears, and the promise and perils of censorship-resistant AI.
• 80:23–82:06: “War footing”: urgency and the need for grassroots resistance.
• 90:43–93:11: The central message: Stop funding your own replacement—build decentralized alternatives.
• 101:19–103:10: Dystopian fears: trajectory to an authoritarian AI future.
• 105:47–107:12: “The darkest scenario…”—the existential risks of inaction.

---

Tone & Language

Throughout, the tone is urgent, direct, reflective, and often polemical. The guests oscillate between optimism in technology’s potential if decentralized, and deep pessimism if left to current economic and political forces. Peter acts as a skeptic, a fellow traveler in the world of decentralization (Bitcoin), and a pressure-tester for the Libermans’ claims, motivations, and protocol design.

---

Core Takeaways

• We are collectively building and paying for centralized AI systems that could render most people obsolete, both economically and politically.
• There is a narrow, rapidly closing window to build decentralized, censorship-resistant AI infrastructure.
• Bitcoin and open-source movements offer a playbook for resistance, but require community, incentives, and collective urgency.
• Unless decentralized alternatives gain traction, AI is likely to entrench new forms of digital authoritarianism—on a timescale of just a few years.
• Personal and economic freedom, the fairness of future societies, and perhaps even the survival of humanity, are at stake.

Transcript

[Daria]

If AI this infrastructure will be controlled by just few corporations or few governments, then we can see really bad scenarios.

[Lieberman]

It's not even a slavery. We are not going to even know it's a slavery. Malicious actor. So someone who actually want to manipulate your opinion or your freedom of thinking around any particular subject, that's it. They own you. We all understand they will replace us and yet we keep paying them to for them to replace us. And in this world we are all losing the jobs and we're all the moving parts.

[Daria]

We maybe have one year, two years and then it will be too late.

[Lieberman]

Currently we are racing towards the wall, towards a really bad outcome. Racing.

[Peter McCormack]

What is the darkest scenario that you almost don't want to say out loud? This show is brought to you by my lead sponsor, Ayron. The AI cloud for the next big thing. Iron builds and operates next generation data centers and delivers cutting edge GPU infrastructure all powered by renewable energy. Now if you need access to scalable GPU clusters or are simply curious about who is powering the future of AI, check out aaron.com to learn more, which is Irena. All right. Morning guys.

[Lieberman]

Good morning.

[Peter McCormack]

Very, very excited to talk to you. We've been planning this a little while. So look, I've been thinking about AI a lot. I'm using it a lot and I'm trying to consider what it means for the future of the planet we live on, especially the world my kids are growing up in. And if AI becomes the infrastructure or part of the infrastructure for everything, electricity, the grid, the Internet, money. What happens to human freedom if we can't live outside of it?

[Daria]

This question, the answer really depends on how we will build this infrastructure. So if this infrastructure will be equally accessible to everyone on Earth, then it's

[Lieberman]

just increasing the freedom. If it's opposite, it's a complete opposite. It's not even a slavery. We are not going to even know

[Daria]

It's a slavery 1984 type of situation. That if AI this infrastructure will be controlled by just few corporations or few governments, then we can see really bad scenarios.

[Lieberman]

Propaganda. Forget about this word. It's not going to be even visible, it's in every answer. We now use the system to be our therapist, to read news through this, to answer the questions. Google, do whatever, like make calculations. All of it is there and the trust is going to increase the less mistakes the system is going to be making. The trust is going to be like enormous 100% trust and therefore whatever actually comes in through the malicious actor. So someone who actually want to Manipulate your opinion or your freedom of thinking around any particular subject that sets your. They own you.

[Peter McCormack]

But if it is part of everything, will people be able to opt out? If for example, they say I don't want to live in this world where I'm constantly at the whim of an AI or talking to an AI, as we've seen with mobile phones, there starting to be a rejection. They want to live in the real world, want to touch grass, want to eat good food. Will you be able to live in this world outside of AI systems?

[Lieberman]

That's a good question.

[Daria]

You will be able. But whether you will be able to be at the same time competitive in the market. In the market. So whether your productivity will be on parity with the productivity of people who use AI, that's a question mark. For some. Like for really artists, small amount of people, it will work. But for most of us it won't.

[Peter McCormack]

Do you imagine there will be any form of money that lives outside of AI?

[Lieberman]

That is really interesting, honestly. If you will go to the extreme, extreme, let's say one corporation, like full extreme, not multiple. One corporation owns all of the AI and now it actually can replace all of the human labor, whatever, like software engineers, designers, management, marketing, all of this. Then the question is like are we all going to lose our jobs? Because the answer can be no. If we all actually keep working with each other, not with the corporation, what changes? Like this corporation produces anything for itself. We don't just don't buy whatever they're doing. Even if they are 10 times more efficient, we still can do stuff in between each other as it is today. Nothing change in this way unless it has been used for control over our decisions. If we can opt out, then we can actually just interact the same market as it used to be, as it is today, except there is another market probably which is like 10 times of what we are doing.

[Peter McCormack]

So the biggest issue here is like with everything, it's the same battle once the Americans kicked the British out of America where they were trying to develop the Constitution. The battle between decentralization and centralization which we saw there with the Federalists, the anti federalists, which we see now with money, whereby we have centralized government money where they print and create whatever they want. And the reaction is Bitcoin AI has, is going to present the same issue of centralization.

[Lieberman]

Not going to, it's presenting as we speak,

[Daria]

you know, like the same way as it happened with Bitcoin. First you will see just few people who realized it who might sound even like super Radical and who will tells us narratives that we need to be careful about, that we need to build alternatives and things like. Like that. But then you will see events after events which will convert more people. Like with Bitcoin, every financial crisis, every time when Federal Reserve printed another like trillion dollars, brought more people into the

[Lieberman]

belief that there need to be an alternative.

[Daria]

And the same we expect will happen with Bitcoin AI in synchronization of AI that every time when we see news, like for example with OpenClaw recently, it's an agent, really popular agent, open source, so anyone can use. Hundreds of thousands of people start to use it. And they were using it with Anthropic cloth, their powerful model. So they were excited to use this model. But then Anthropic just shut it down. Shut down only for users of OpenClaw. And it was a big effort for the entire community to try to switch off and try to build everything which is needed to switch off the model. So the access to the model and access to AI is already right now is visible that we shouldn't take it for granted. We shouldn't expect that those models will be as available for you as they are right now. And when you think about it, you can see how you already right now making yourself more and more dependent in your line of work, whether you an engineer or you are a business owner or you are journalist. And in that case, every time when people will see that their freedom is challenged, we expect more and more people to look for the alternative.

[Peter McCormack]

I think the word slave is a word we have to be very careful. It is a loaded term. But I'm going to tell you a little story. Recently I have been weaning myself off my mobile phone because it was explained to me once, this is like a parasite. And I notice in life, wandering around, you see people with their phones all the time walking down the street, on the buses, on the trains, they're just glued to their phone. So I deleted in one day around, I can't remember the number. 130, 150 apps. And I've been deleting about one app a day every day consistently. And I'm trying to get my phone to the point where it's only a tool for booking things, flights, paying for parking, banking, but no distraction apps. I want to get away from it. I'm trying to wean myself off it. But conversely, I'm going deep with AI. I am creating AI infrastructures for my companies where it helps me prepare things, get jobs done. And I'm wondering at what point with AI do I go from it being a tool for me? Do I be a slave to it?

[Lieberman]

We have to be careful with this word slave. I agree with you. Some say from what we see, people around us like people in the entrepreneur's community, people with multi hundred million dollars billion dollars companies started working more because of AI. They just cannot stop. Because it's very addictive in a sense because you just say and stuff happens, like you just write down, type down your commands and the stuff is happening like software being built, marketing campaigns being prepared, emails was written, GitHub commits was done for you. So it's in a way really addictive, like how much stuff you can do with this and how this is like an opposite from slavery. In a sense you are a slave master.

[Daria]

But.

[Peter McCormack]

But do you know, that's interesting you should say that. I discussed this with somebody the other day and I said I think I'm suffering from AI exhaustion.

[Lieberman]

Exactly.

[Peter McCormack]

For two reasons. Because it's made me more productive and enabled me to do more. I'm actually working more more. But secondly, trying to keep up with the developments in AI and the changes is close to impossible. And so I'm starting to feel like, do I want to go down this rabbit hole? Do I want to be this guy? Because this is exhausting.

[Daria]

The only way to keep up with AI right now is to be unemployed,

[Lieberman]

literally. This has to be your job today to keep up with AI. But, but at the same time, what's interesting is I'm not sure that is that needed. What if you will skip couple months and then come back and look at what's happening right now. Like all of the model is going to be two upgrades further, some additional tool will appear. This will be replaced with that and stuff like this. And the honest answer likely is that you are not going to be lagging behind. It just it will take you some time to get in into the new tooling set and just continue. So I think that we are overexcited right now, all of us, because like the magic is happening right in front of us. But at the same time we are addicted to, to this and we are. And the anxiety of like the FOMO of not being up to date with the latest tools when your friends, peers, competitors or whoever it is, or your girlfriend is actually doing like knowing more than you in the sense of what's happening in AI, this is real problem right now.

[Daria]

But that that's create the dependency and dependency is already much closer to the slavery. Yes. Because if your livelihood, whether you can earn. Whether you can actually be productive in society depends on that. Tools, then you can adjust, stop using them.

[Peter McCormack]

Well, if you think about how we've aggregated things with the Internet, we have Amazon, which has been very useful as a tool, but terrible for bookstores. Bookstore owners, it's centralized everything to Jeff Bezos. If we think what's happened with things like Uber and UberEats and Deliveroo, all these companies have created this gig economy where people have essentially the jobs at the bottom rung of the ladder trying to just make enough to get by. But they're the kind of upwards mobility there's very difficult. With a bookstore, you could work on your bookstore, you could, you could grow it, you could maybe open a second bookstore and a third bookstore. You know, if you have, I don't know, if you had a coffee shop, you could have a second and a third. But we centralized everything around technology to, to make that upwards mobility very difficult. But we have created a lot of super wealthy people at the top. And yes, lots of other jobs have been created. But you know, what does that do for the soul of somebody wants to have a great life? Well, can I just finish the point? So what I, what I'm thinking now with AI, I will listen to something like the all in podcast. Sax and Jamath whoever talk about the power of AI to see Mark Henderson talk about the power of AI. But what I've noticed is, and I take this from an Eric Weinstein tweet, so I'm not, it's not wholly mine, this is really his idea he implanted in me. But these LLMs are sucking up all the innovation of human history and they're centralizing the upside to a fewest group of people. And if the AI starts taking the jobs, these people, they don't have any skin in the downside of AI. And at what point does the AI just start distributing jobs a bit like a gig economy and we are functionally slaves to the system.

[Lieberman]

This is the scariest. Like, this is the way we see this. We only have two scenarios. One is the one which you're describing, all of us are losing the jobs. Some specific corporation or a legal poly of like five of them who are all basically the same, producing the same product. It's just verticalization of everything in, within the specific ecosystem, whether It's Apple ecosystem, OpenAI ecosystem, Google ecosystem or whatever is all the same. Verticalized from the top, like your intelligence to the bottom, like your toilet paper, like all of it centralized in one vertical. And in this world, we are all losing the jobs and we're all. The moving parts. You're like ants who are like. Yes, exactly. So like I was trying to find the analogy, but the ants of just bring the stuff from here to here. Like a physical movers, physical world movers for the.

[Peter McCormack]

Until the robots kick in.

[Lieberman]

Until the robots kicks in. Exactly. And this is one world, but the other world, the other world, the alternative one, the beautiful one, is in which each of us have the robot and each of us have the same equal access to, to the superintelligence. And if it's equally distributed, then each of us just empowered with a tool. It's a new productivity tool. Each will be super productive to the extent which we want to be productive. And we will just continue the same market in the same economy unless we reach. Not unless, but until up until the moment we reached the complete abundance when the robot and the intelligence which I own can produce everything I need in my life for myself. And now the only thing which is outside of the scope is relationship with other people. And we'll just keep having the relationship. Maybe they're going to be different and weird, not as we expect them to be, not as we have them today, but in this world, like the superabundance before the superabundance, we can move forward with the equal access to this particular tool. And the more equal it is, the less disruption we will see. And this is all about access.

[Peter McCormack]

Yeah. So what I'm thinking is there aren't enough people thinking about this as a problem. There's no incentive for the people who run the all in podcast to think about this as a problem because they are Silicon Valley investors who have benefited from centralization. And in some ways centralization has won over the last 100, 200 years. The Federalists won in the US the Anti Federalists lost. But they were correct because look what's happened to the money. The money's been centralized. Bitcoin has a such a compelling story and everybody who's seeing everything go up in price, all the things again, expensive and they're losing to inflation, they're still not coming to bitcoin. And so in the world of AI where we have these super amazing tools, these incredible LLMs that do things that are like magic to us, how do we get people to even start thinking about. There must be an alternative, a decentralized

[Lieberman]

version A, we do have the current situation when people getting more and more scared of where it is moving. And this is important. I mean, I hate the real development to be happening within the Humanities, history, from the perspective of fear, like, I hate this, honestly, that's why I want to present the positive picture. And rather all of us actually strive for having this better, positive future. But the fear is real. I mean, the problem is real and we can feel it right now more and more. And we will feel it more and more within the next two years. I mean, it's going to be super fast.

[Daria]

So currently, for example, almost every week you see news of additional layoffs, another 10,000 people. And people should stop for a moment and think about it. If the claim that these tools increase productivity is right, is correct, then why would you fly off people in opposite? You can do more, you can build more, you can accelerate the economy. So why would you layoff people?

[Peter McCormack]

Well, you need different kinds of people,

[Daria]

though in this case, you still should have more hiring than they layoff, which we don't see. The challenge here is that these centralized companies, they originate, when they start, like with Amazon, they bring us really like, great idea, for example, that the goods can be delivered to your home just in like 5, 15 minutes, just one click and we all buy in. In this narrative, we really want this world to happen. And then they change the policies. For example, Amazon was like, web platform, we're just selling goods. But then they're looking at the data that you buy into goods and then they offer you generics, which were much worse. And they actually eventually more expensive than the ones which were there on the market. Because when they control the platform, the platform they can actually then manipulate. The same happened with Google. Google promised, I think it was 2003 or 2004, that the ads will never be in the main feed, only in the search feed, in the search feed, only at the side of the screen.

[Peter McCormack]

Yeah, but they also promised us they wouldn't be evil. And then they said, oh, well, we're not going to say that anymore. It's like, what do you mean when you say you're not going to say you're not going to be evil anymore? Are you saying we're going to be evil?

[Daria]

So this is a problem. So when we have decentralized protocols, their rules are encoded and they never change. And that's what creates certainty, that we can all trust the protocol and we all can lean in and be passionate about them. But with those centralized organizations, they don't believe that their statements, which they made early on, should be persistent. Persistent.

[Peter McCormack]

Do you think they're gaslighting us basically, because there is so much upside to them being right.

[Lieberman]

What's interesting, what's interesting. Who, who are they? That's the big question. If you. Sam Dario INDIVIDUAL and then the corporations like Google or Microsoft, it's thousands of people making their own decision for their own benefit and like having the bonus or, or stuff like this. And then there is like a layer of investors like shareholders who are mainly some funds, like faceless funds. There is no one actual skin in the game in there. They're just making money for someone else. It's a chain of making money for someone else or for your own. And whatever is happening there, like profit, profit, profit. There's nothing bad in the profit unless you actually become the monopoly in specific sector. And this monopolization leads to the degradation of the service sucking more money. Like we can see this with Netflix, for example. It's just like the price is just going up and up and up and up. And the sort of claim is, oh, this is inflation, but it's not. I mean you can put these numbers

[Peter McCormack]

together more so but how many times do you switch on Netflix now and you cannot find something you want to watch.

[Lieberman]

Exactly, exactly. Something which used to be that used

[Daria]

to spend like hundreds of millions of dollars to House of Cards type of series in. Not anymore. Moreover, now they have more users, so they have digital product which are. Which should be like the cost should be distributed among more users. So it's actually the cost should be lower, the price should be lower.

[Peter McCormack]

But that is inflation. That is inflation.

[Daria]

They're growing higher, like they're growing faster than inflation. If you will just compare their pricing points. But still they increase their audience multiple times. So the cost per user was supposed to go down multiple times and we don't see that. So our argument that all of these companies, they start with great narrative, which, not the bad one, we really want to have it in the world. But then they don't have any incentive to stay with this narrative. And that's what we expect with AI as well.

[Peter McCormack]

All right, let's talk to you about my sponsor, Leden. Now, if you're borrowing against your Bitcoin, there's one thing that matters more than anything else. Is it actually safe? Well, Leden have just dropped their lowest rates ever. But more importantly, they haven't changed the thing that matters most. Your Bitcoin is never lent out, so they're not chasing yield. There's no hidden risk and there's no rehypothecation. Just collateralized loans done properly. Now They've done over $10 billion in loans and they've done this through bull markets, through bear markets and everything in between. And they've done it without ever losing any client assets. So now you get lower rates, which means the bigger the loan, the lower the rate and full transparency before you apply. There's no monthly payments or early repayment penalties, and they give you tools to stay in control from auto top ups to LTV alerts. So you're not choosing between a good rate and safety anymore, because with Leden you get both. Now you can check out your rate using the calculator at Leden IO Peter Mike Green, who's been on my podcast a couple of times, he's an investor. He talks about monopolies as one of the biggest issues in the economy and that we have to have competition. If we don't have competition, this will lead to products becoming worse and more expensive.

[Daria]

Moreover, a month ago there was this podcast of Darkesh with Daria. And at some moment Darkesh asked him but he believed that they will still be working with through API model when people pay for AI tokens. And Daria first said that this model likely will persist. Will persist. And then. But it's always but. And then the but was that he said but not all AI tokens work the same. And then he made this example that when AI helps you with your MacBook, probably you will be ready to pay less, but if AI helps you with your health, you will be ready to pay much more.

[Lieberman]

Which, which in a. From the perspective of let's say electricity, we want this to be. We want AI to be as accessible and as available and as cheap as electricity. But from the perspective of electricity, 1 kilowatt of energy providing you with the. The light in your house and it's just like you can, you. You can use candle, whatever, like you pay less and. And then another kilowatt of power is your. The tooling for the surgery while you are being operated on open heart. And this is kilowatt. If there is a difference in the price of those kilowatts of power, then you will be ready to pay more for the one which is your life as at stake, rather than the one which is basically having more light in the house.

[Peter McCormack]

This is the scenario where we become slaves to the system. We no longer are humans. We are subjects of an AI system.

[Daria]

Yes, that's where we see the difference. So the centralization when the access is not guaranteed for all of us by protocols, then we definitely will get to this future. Unfortunately, the leaders of the industry, they even already don't hide this. They publicly state that so because right now raising Additional capital and preparing for IPOs. That's why they pitch to investors. And for investors, the idea is that AI tokens can be sold even with a higher tag because of some value which they bring. Additional value definitely resonate well.

[Peter McCormack]

They're in a tough spot themselves. They're all competing with this race for super intelligence, which I still don't 100% know what that means, but I kind of know what it means. But they're all in this race for the super intelligence. Who's going to hit it first and what does that actually mean for them. And they're also in a tough spot because of the cost of running. The power for these LLMs is huge. But the revenues aren't just there yet, they're growing. So they're in a. They have no incentive because of the incentives of becoming the first this spot. So why would they care about the impact on all us peasants?

[Lieberman]

They will. They will. They will not. They won't. I mean, there's a simple question.

[Daria]

This was our personal conclusion. You know, since 2014 we were parts of these groups who discussed potential bad scenarios of artificial intelligence. And there were many different point of views of what are the actual bad scenarios, Whether it's like Terminator type of scenario or whether it's just doomsday where we all will be destroyed. But our case was always that most of the worst scenarios, they start from us having AI being controlled by just a few companies. And then when you think through those scenarios, you definitely see that this is inevitable. That even if we believe that the original founders really good intentions and really they really want a better. Future for all of us, but the way they structured will eventually steal, will get us to this bad scenarios.

[Lieberman]

Quite pessimistic. Right.

[Peter McCormack]

Well, so I've done a few shows on AI now and it's often from a place of fear or pessimism. But I always ask the same question, is are you a fan of AI? Do you like it?

[Lieberman]

Honestly? We are. Honestly, we feel that probably AI and the potential future superintelligence is likely the only way out from the shithole we're like heading towards, only increasing the speed with which we're heading there. Because AI in a sense might, if it will be independent in its structure and in its infrastructure as well. AI can enable immutable bureaucrats, immutable. Middleman in a sense, if we'll look

[Peter McCormack]

at this, get rid of the managerial

[Lieberman]

state, getting rid of any state which can be corrupted, any layer which can be corrupted and which has bias or its own interest.

[Peter McCormack]

So Your interest in the idea of

[Lieberman]

President AI Our interest in the AI is more like, you know, right now AI is being used even to design new rocket engines. And when the rocket engine is being designed by AI, the result is something which we've never even thought before. I mean, the shape, the curvature, the idea behind what is happening there, it's alien structure.

[Peter McCormack]

Leap 71.

[Lieberman]

Leap 71. Probably, yes.

[Peter McCormack]

Okay. Wow, that looks like an alien spacecraft.

[Lieberman]

Exactly, exactly. So, like think about this. So when you set up a deterministic AI model and you set a specific goal for this model, and the model is designed and created for solving a specific problem, then the models can actually come up because of the ability of the model to browse through millions and billions of variations, which we would have spent thousands of years to do the same in the lab, in the researcher lab. Then we can come up with something which looks completely not what we expected, but it actually does the job. It actually gives you the power, the temperature, the, the curvature. Like everything you need from the physical perspective you will get. So from this perspective, do we need a president AI or do we need the AI to design a system for the humanity to achieve the most efficient state of self coordination. But for that, like a completely decentralized self coordination when no one have the lever to get the power to even became the dictator benevolent distribution of power. The distribution of power, the most efficient system for everyone to achieve in their most potential, the highest potential. Like this. So we can create the AI model which will actually inevitably design the system which will increase freedom, rather than makes us all slaves. Because current movement towards the slavery is not the AI itself. It's the design of the, of the corporations and the design of the political

[Peter McCormack]

systems and the financial system and the

[Lieberman]

financial system, political and financial, they're like right now together. Like financial, political system is the same creature today.

[Peter McCormack]

Well, so the risk and the problem we face is we live in a financial system where we need money to survive. That is what it is. But it's a system which extracts upwards.

[Daria]

Yes.

[Peter McCormack]

Governments always spend too much money and banks create money out of thin air as it creates inflationary environment. The assets inflation late and your wages don't keep up. And that's under a lot of pressure at the moment. I mean, I think you see on social media, the amount of criticism of billionaires and large corporations is bringing in this kind of new dawn of communism. People are just nudging us towards communism, which we know like fucks everything up anyway.

[Lieberman]

Yes, yes, yes.

[Peter McCormack]

But you can understand why people are thinking like this, because it's not like in a recession where you lose your job and you hope for the economy to recover and you come back.

[Lieberman]

The economy is flourishing, someone is getting billions of dollars, and like the bottom is actually getting poorer and poorer.

[Peter McCormack]

But those jobs that are going are never coming back. You may have gone to university and trained for, you know, whatever many years as a lawyer. And you know, once that job goes, there's no lawyer job for you left. Maybe the guy who can coordinate the lawyer agents gets a job. Those jobs are not coming back. And maybe you're 35 years old, 40 years old, you got two kids, and maybe your wife works in the creative industry and her job's gone. It's like suddenly you've gone from a nice life, holidays, kids in private school, and you can't afford shit now. Like, what is the plan for that? And the only answers we're hearing is ubi. So like, we. It's not only that AI brings this thread, we have to rethink the entire system so people can live and flourish.

[Daria]

But ubi, UBI is, from our perspective, is really bad choice. Because no One says that UBI will be like $10,000 a month.

[Peter McCormack]

Well, he almost did. Universal high income.

[Daria]

He said high income. Yes, but, but, but unfortunately the problem is that someone, if someone decides who will get get the UBI and how

[Lieberman]

much, that someone is deciding what qualities you should be, how you should behave, to whom, obey and do stuff like what law to follow.

[Peter McCormack]

They have power.

[Lieberman]

They have power over you. And now they dictate everything you can do and cannot do. And unless you do this, you are dismissed of the ubi. And if you do this, you are getting your ubi. Like, this is a shitty world. None of us want to live in the shitty world.

[Daria]

So we would prefer the world which Daniel mentioned, when every human on earth have their own robots which can create everything which I need to consume.

[Lieberman]

And think about this, in this case,

[Daria]

it's much better than ubi. Like, you actually have not universal basic income, you have universal basic access to the.

[Peter McCormack]

You have sovereignty.

[Lieberman]

Yeah, sovereignty. So what's interesting, think about this. Today I would say probably over a billion people, like a billion people in the world will be, will have enough wealth, enough capital, enough savings to buy a robot which would cost $20,000. Like Elon said, like it's going to get to $20,000. It will get to the 20, $20,000. Why? Because the motorcycle is getting there. And the motorcycle has more physical material in it than a future robot. The future Robot will weigh 50, 60 kilos. The motorcycle is like couple hundred kilos. When you think about the through this perspective, the motorcycle is different from the robot only because of the design and the brain. And design is a file, is a CAD file and the brain is AI. So there's difference in tells you that the robot can cost 10, 20,000 dollars. If the robot can cost 10, 20,000 Dollars. And today people can afford it still, people can still afford it?

[Peter McCormack]

Well, it depends. I mean if I can afford it, you guys can probably afford it.

[Lieberman]

A lot of people afford a car. For as many people who can afford a car today, they will be able to afford a robot.

[Peter McCormack]

They might be able to afford the second hand robot, the used robot. When I get the new one, you get the new one. We sell our old robot.

[Lieberman]

And if your robot can do the

[Peter McCormack]

job, I wonder if you're a robot, become like a dog that you won't want to sell it because you become attacked.

[Lieberman]

It's your, it's your, it's your second half. In a sense you're like part of you now. This is you with your 24 hours a day. Eight you're working eight you're sleeping, eight you're doing stuff for yourself and your robot, which is only doing 24 of working, let's say, and which actually increases your productivity four times at least. And this is the world which, in which we can live.

[Peter McCormack]

Hold on, what happens to the global wealth divide there? Because yes, in America and in the UK at the moment, although we're heading third world, what if you're in Ghana or Kenya, not everyone will be able to afford a twenty thousand dollar robot.

[Lieberman]

Yeah, but, but think about the Thursday's perspective. If we increased our productivity four times, if the developed world increased our productivity four times, can we afford to build additional robots and then be sending them to Sagana? And what's interesting, we can even sell it for them for the future share of the robot's productivity. So we say yes, we build your robot. We're not donating you the robot, we're leasing you the robot for 20% of its labor coming back to us. And 80% of the labor is yours for you to build your life. This is a clear passage towards what we can imagine as a complete abundance for the entire humanity. Unless we do the war with it. We do the fight for power with this. We do the concentration of monopolization, enslaving nations and stuff like this. There are two scenarios. They are. This one is inevitable in a sense. Unless we do something to Move towards a different one.

[Peter McCormack]

But if you think about the history of humans, we've always taken the wrong path. We've always taken the path of centralization. What can I have? Quick, what can I have now? What's easiest? And there's always been psychopathic, but we kind of have.

[Daria]

You're right. You're right that there are a lot of reasons to be pessimistic. At the same time, we do have several good examples. Okay, we have open source operating system, which dominates in the world of server.

[Lieberman]

Everything we know about Internet and everything we know about protocols today, everything is open source.

[Peter McCormack]

Why did that work?

[Lieberman]

I would say because of a specific small group of people, like not more than a million people in the world, highly educated, highly motivated, with a high eq, like an iq, together, decided that they want to live in a different world. And they committed and contributed towards the common prosperity and common good. And we can do this now because another example is bitcoin. Few people understand. Yes, bitcoin is right now, there's a lot of billionaires in the bitcoin ecosystem. Someone made a lot of money. Someone is not making any money. But bitcoin, if it would have worked as a actual transactional currency. That's the problem with Bitcoin right now, that it's only doing seven transactions a second, which is not.

[Peter McCormack]

That's not entirely true.

[Lieberman]

Not entirely true. I agree with you. And we can actually upgrade the system to get to the point when it can become the currency. I completely can see that working. But what happened with Bitcoin is that over $300 billion was eventually invested towards the infrastructure of bitcoin. Like, that's a lot of money. And it was not the largest corporations of the world or even governments. It was grassrooted community of people, millions of people, building smaller data centers, filling them with asics, developing new chip designs, building new electric electrical systems, building new data centers, all of the infrastructure for bitcoin.

[Peter McCormack]

Hold on. But that was economic incentive. There is the other side of that, which is the developers. And there are plenty of developers, I mean, I know some of them who could have gone to Google and got paid $300,000 a year, $400,000 a year, and stock options and become millionaires. And they chose to work on bitcoin funded by individuals. There's groups of individuals who funded. There's. I funded bitcoin developers.

[Lieberman]

So that's why I'm saying like you

[Peter McCormack]

need a combination of economic incentives and then people with the right personal incentives.

[Lieberman]

You need to build a proper Market insensitive. Economic market insensitive. Not many. Not monetization and not gatekeeping. Like not gate, like a completely open ecosystem which by design and bitcoin showed it drives the price of the service down. What's interesting, the service in Bitcoin, in a sense, it's a strange concept because like the bitcoin goes in price itself. The Bitcoin. So it looks like it's getting more expensive, but not because the terra hashes which have been produced by this system are getting cheaper, cheaper and cheaper. We just produce them in an amount which is enormous amount of mathematical operations produced by the system per watt of electricity is getting cheaper and more abundant. If we could have used them.

[Peter McCormack]

I'm going to come back to you on this seven transactions a second. Just because I want people to understand that is not the reason Bitcoin is not being used as a currency. We are also talking about the base chain. There is the lightning network with its imperfections. But if you have a lightning wallet now, you and I, we can do the more than seven transactions a second. So it is possible to use it as a currency. The reason it's not being used as a currency is it's become a victim of its own success in that in a world where, you know, the pound is dying, the dollar is dying, the fiat currencies are dying due to inflation, you don't want to use your bitcoin, you want to spend your dollars. The last thing I ever want to do, dude, I sold some bitcoin to pay for the deposit on a house. If I'd had just held on that bitcoin for three years, I could have bought the house outright. And you go through those experiences, you go, oh, I'm just going to hold Bitcoin for now. Because what we're doing is we're swapping currencies. So why spend it? The same happened in Venezuela. People would, you know, borrow Venezuelan. What is it? The Venezuelan. The currency was. And they would buy dollars, right? That's what they would do. They are shorting the currency. It's what Michael Saylor's doing. Michael Saylor is shorting the dollar because he knows it's screwed. So bitcoin can be a currency. We just have to kill off the fiat currency first. I just want to correct that one.

[Daria]

Yes. And that's true. And we just need to. We believe and we actually see a really optimistic sign in that because what Daniel mentioned that the price of the terrahash decreased. It's not just decreased, it decreased 300,000

[Lieberman]

times from the very beginning to where we are today.

[Daria]

So it used to that you would spend 5 million joules joules per terahash. Now it's just 15.

[Peter McCormack]

So you're saying the same can be done with tokens?

[Daria]

We believe that with AI will happen the same. You're totally right. But it will happen in the same way only if we'll have the same type of structure which constantly drives the price down.

[Lieberman]

Incentivize building infrastructure more efficient. More efficient. Whoever is building more efficient infrastructure, getting more coins.

[Peter McCormack]

You need the infrastructure built efficiently. But you need a committed and dedicated passionate community who say no, this is the world I want to live in. I don't want to live in anthropic and grok and you know, OpenAI. I want to live in this different world.

[Daria]

Yes, and we believe that the same kind of trajectory can be expected. That first it will be hundreds of thousands of people, but then it will be millions of people, then dozens of millions of people. Because every time when we will see another layoff, or every time when we'll see see another gatekeeping of the access to the models, more people will convert.

[Peter McCormack]

Well, there's another point to this as well. So I am a bitcoiner for two reasons. There's the financial incentive reason and the second one. I like anything that takes power away from the government. I love it. I have my bitcoin. The government can't touch it, they don't know what I've got. Fuck those guys, because they will steal everything when they can. So I love it. It's like an anti establishment movement. It is a grassroots give power back to the people movement. If there is an AI movement which is grassroots, which is saying no to the big corporations, no to government, we're going to do our own thing. Fuck you, I'm in.

[Daria]

Yes, that's what we expect. And moreover, bitcoin also showed that it will not only drive price down, but also the infrastructure itself can be built much larger than we can even expect. Because bitcoin today is comprised of Data centers with 23 gigawatts draining 23 gigawatts of power.

[Lieberman]

It's more than Microsoft, Amazon, Google, OpenAI, meta combined.

[Daria]

So imagine. So when we think about it, whether AI can be decentralized, whether actually the community of passion people can build something over a similar scale. Bitcoin actually proved that. Yes, within the same time frame, bitcoin exists two years less than cloud exists. But still bitcoin achieved more in terms of really deploying the infrastructure than the entire silicon Valley so financially sensitive and

[Lieberman]

people who are dedicated to the mission, to the belief that the system need to change.

[Peter McCormack]

But does this community exist now? Because I can tell you the Bitcoin community, the pockets of resistance, where they exist, I know who the people are. It's a good community, gone through some challenges I think over the last couple of years as Bitcoin's been financialized and adopted by Wall street, etc. And governments. But it's still a decentralized protocol. It's still secure. Like I still hold my private keys, you still can't, you still can't get them off me, right?

[Daria]

Yes.

[Peter McCormack]

Is there a community around this right now?

[Lieberman]

It is happening as we speak. So we launched our protocol, the protocol which we built, what is it? Seven, eight months ago, and it already grew to almost half a percent of this of the GPU size of OpenAI. Completely decentralized. There is no foundation or company, any structure behind it. It's all the community of people coding, adding the hardware, adding more GPUs and then preparing the system to be able to be used by anyone in the world.

[Peter McCormack]

Is it its own LLM currently?

[Lieberman]

Currently it uses anything which is available as open. Wait, open source, a bit like Venice or other one.

[Peter McCormack]

You can just select the model you want.

[Lieberman]

Yes, you just select the model which you want from the Chinese or available like Gamma or Llama or Deep Seq or Quan, whatever is available out there. But the whole idea is actually that the protocol is designed so that in the future, not just in the future, but it designed so that 20% of the network's power reserved for training future open source models. So the service which has been sold for their tokens is as a utility tokens is inference. But the training of the new models is not sold, is not for sale. The training of the new model is free for everyone to use to train. Open source model, completely open, open from the data sets to the training software to the final model.

[Peter McCormack]

So you can have a completely decentralized

[Daria]

LLM, absolutely immutable LLM. You know, like we in crypto world, we had two major revolutions. First was with Bitcoin, with the introduction of immutable ledger. So as you mentioned, you send a transaction, no one can stop it. Yes, government can stop it. Like no one can stop it. It will be there. Then with Ethereum, introduction of immutable contracts. So we agree about something, there is a contract and no one can change will be executed still it will be executed exactly as it's written on chain.

[Lieberman]

Have some faith.

[Peter McCormack]

It's still a shitcoin.

[Lieberman]

The shitcoin, not shitcoin. The idea was.

[Peter McCormack]

Yes, I understand the idea that we

[Lieberman]

need to introduce immutable concepts, simple script being executed immutably.

[Peter McCormack]

Yes, I understand.

[Daria]

It's the same we need with AI when we have a model, you send your input tokens, you send the request and you know for sure that the output which you're getting is exactly how the model is defined. Because in reality most of the censorship right now happens not inside of the model. Most of the censorship happens through the pre prompts that's there, which we don't really see that. All these companies, they add in some layers of filters, of commands which the model obey, don't control.

[Lieberman]

You remember when Deep SEQ was only released, a lot of people started asking harsh questions, specifically in the Chinese policy, like who owns Taiwan? And then the model, when you send the request directly to the website of Deep Seq, the model will say like we people of China are committed. Literally, like, literally this would be the answer. While at the same time, if you actually run the model on your own local instance, the answer is not like that.

[Daria]

The answer was it's free, like it should be free, and things like that.

[Peter McCormack]

Well, that's like Google. When you would ask Google within, inside the great firewall of China exactly about Tiananmen Square, it would just show you some pictures. It's a lovely place, nice part of China. You ask here, it's like, yeah, it's where they fucked the people.

[Lieberman]

Exactly. So that's the whole point. So it's not the model itself because the model is trained on data, vast amount of data. All of the humanities knowledge which was collected for thousands of years now used to train this two terabyte model which can fit in the memory of one server. This is like the models are tiny in a grand scale, but the model itself is sort of unbiased unless there was a lot of biased input in the data set. But the model itself usually really unbiased and it's a pre context which then filters everything the model answered to you or filter everything you asked through the model to the model before you even get in the result. So the openness needs to be on all layers and the immutability needs to be there as well. If we actually want to build this. Immutable bureaucrats, immutable middlemen.

[Peter McCormack]

Okay, then I have to ask you, if you're building this and you're right and it's successful, does it make you rich?

[Daria]

In our case, we have some coins,

[Lieberman]

we have some Coins.

[Peter McCormack]

What do you mean by coins?

[Lieberman]

Coins. We have coins. So the system which we built have a limited amount of coins, which is a billion coins.

[Peter McCormack]

What is this? A token.

[Lieberman]

It's a token.

[Peter McCormack]

And is this like a crypto token or is this a token that is used within the system?

[Lieberman]

That's why I use the word coin.

[Daria]

Because otherwise in AI world, tokens are used in a completely different.

[Peter McCormack]

So this is like a crypto token.

[Lieberman]

This is like a crypto token. A currency with a limited amount of coins in circulation always will be not more than a billion coins, but fixed. It's fixed. And you can pay for the inference to be done for you with this coin.

[Peter McCormack]

Or you need that coin.

[Lieberman]

You need that coin to actually use the API of the decentralized AI.

[Peter McCormack]

And could you have built it in a way that it didn't need coin? Say I could have paid with Bitcoin.

[Lieberman]

Yes, we could. But the reason why we did this was specifically to create this financial insensitive, this economical reason for people to rush in and add hardware into the system. If we want to build an alternative to OpenAI and anthropic and Google, this need to absorb hundreds of billions of dollars of investment into infrastructure.

[Peter McCormack]

So you have to incentivize the early people. Yes, a bit like with Bitcoin in the early days, if you mind it, you got 50 Bitcoin a block and then it. Okay, so can I ask a question?

[Lieberman]

Sure.

[Peter McCormack]

So you have the coins, but who owns the company?

[Lieberman]

There's no company.

[Peter McCormack]

There's no company. Okay, good start.

[Lieberman]

No company and no foundation, no legal entity behind it. They're completely open source, completely decentralized. We're not even mining ourselves. Specifically, we found eight independent miners, Bitcoin people, people with experience in cryptocurrencies, convince them that this feature need to exist. They launched the network. The genesis was them.

[Peter McCormack]

Okay, can I ask, what percentage of the coins do you guys have?

[Lieberman]

The four of the Liebermans owns 10% of the coins.

[Peter McCormack]

Okay, so.

[Lieberman]

So two and a half for each of the four siblings who was spending last two years developing this protocol.

[Peter McCormack]

And that's a brother and a sister, right?

[Lieberman]

Two of us and the two sisters.

[Peter McCormack]

Oh, two sisters.

[Lieberman]

Two sisters.

[Peter McCormack]

Where are they? Why are they not here?

[Lieberman]

They need to be here. I agree with you.

[Peter McCormack]

Okay, so. So now I'm not going to grow you hard on this. The great thing about Bitcoin is Satoshi Nakamoto who. We still don't know who he. They. She is.

[Lieberman]

They.

[Peter McCormack]

They maybe talk about that offline, but we don't know who they are. And as far as we know, he had no financial incentive and didn't want the quote coins. So if you're building a big decentralized network, people just have to be aware that you have an incentive to promote this and you have an incentive to be critical of the centralized systems. Now luckily I agree with you anyway because I'm a decentralization guy. I'm like fuck the government. Anything away from that is great, but people should be aware of that. So you should be challenged on that.

[Lieberman]

And we are challenged on that constantly. Our point, everyone's like wait, Satoshi didn't take anything for themselves. And at the same time Satoshi was almost the only like there was few more people who was mining for the first year or so. And eventually on the satoshi's wallet collected something up to 10% of all the bitcoins ever mined. In this sense, Satoshi actually got the financial incentive never used. And this is sort of the same with us. We even the coins which we got, we even put them in a four year vesting so they're not affecting current agents of the coins with the miners like coins which been mined daily.

[Peter McCormack]

This is your reward for saving the world from digital slavery. But if you do you you might be four of the wealthiest people in the world maybe.

[Lieberman]

And at the same time this is not only the reward, but this is only also a resource for us to compute to keep actually resistance alive.

[Daria]

Because if you have this tokens you can use them in two different ways. Two different ways you can sell them, but you also can use them to get compute.

[Peter McCormack]

Is there an open market for all these coins at the moment?

[Lieberman]

Only created because we're U.S. citizens. We want this to be completely accepted as a cryptocurrency completely decentralized. We're not even touching the aspect of this coins to be sold. But in the past half a year community built at least three or four dexes already DEXS and OTC platforms to start trading the coins. So there are availability.

[Daria]

So the original. So there was several good attempts before to build a decentralized but like bittensor or but problems that they all were built on the kind of. On the premise premise of proof of stake. They all get security through proof of stakes. And maybe it's good for like financial deals for years.

[Peter McCormack]

Proof of stake centralizers.

[Lieberman]

Yes.

[Daria]

But the problem of proof of stake that all incentives all the reward just goes towards capital but not towards the people who build infrastructures. So eventually even though these projects are great and they Achieved good results in terms of market value, but in terms of GPUs actually connected to the network, they weren't so great.

[Lieberman]

So in our case, imagine like I'm speaking about this immutable bureaucrats of the future. If we want to build a completely different future, we need a lot of tokens, AI tokens to be spent on something which is not going to be for the financial interest, something which is going to be for the good of the humanity. And we want to retain this 10% of our coins for our personal usage or for the usage which we foresee that in the future we will need these tokens to run most powerful AI agents which will be bringing a different point of view on self coordination of

[Peter McCormack]

brings a certain amount of responsibility for you guys. And it will put a different lens on this for people. But.

[Daria]

Yes, but at the same time all the tokens which are issued every day, they issued toward miners. In our case, through proof of work system. We kind of went back to the roots to what bitcoin introduced. Because we believe that Bitcoin actually showed us the the path of how we actually can build infrastructure. Infrastructure and the only way to build it is proof of work when people who can create better equipment, better chips. Yes. We used to have the state where Nvidia was a dominant player on bitcoin bitcoin mining in 2012. And then in 2013 it changed because the convenience of engineers, of tinkerers, they actually found a way how to make better hardware, more efficient hardware.

[Lieberman]

Some might say it's cheating, but in a sense it actually proved the whole point. The insensitive allowed, not the whole point. I doubt that Satoshi ever actually dreamed for this to end up that way. And some people might say this is bad because we're wasting a lot of electricity on the same Bitcoin. The bitcoin didn't change. It's like we waste more and more electricity on that, but at the same time we believe that this is actually coincidence or something, whatever. Or the world needed this and that's why it appeared.

[Peter McCormack]

Yeah, well, waste is subjective.

[Lieberman]

Waste is subjective.

[Peter McCormack]

Yeah, I think, I don't think bitcoin wastes any energy. It just becomes more secure.

[Lieberman]

So what's interesting, like for example, some might say that Instagram is waste of human beings time. But then Instagram allowed to create all these billions and trillions of pictures which then was used to train AI to recognize the world. So was it waste or was it not waste? All of the information which we offloaded into the Internet and made Public now used to train AI, which if we drive it correctly, direct it correctly, we can achieve abundance for the humanity and we can achieve a higher freedom, higher state of freedom, a state of higher freedom in the world, around the world, than we ever even can imagine.

[Peter McCormack]

I want to talk to you about one of my sponsors, Incogni. And that means we're going to talk about the weird world of spam. And I don't just mean those spam emails that you get day after day from companies you never heard of and companies you've never signed up to. I'm also talking about those spam phone calls you get from those people who seem to know a little bit too much about you, trying to get your bank details. It's all a bit creepy right now. This all comes from the world of data brokerage. There are companies out there collecting your data, building profiles and sending that data to anyone who wants it. Which is why when one of those scammers phones you up, they seem to know everything about you. Now, I've tried, I've tried myself to get off these lists, try to get off the phone lists, try to get off the email list. I unsubscribe from every one of these emails that comes in this game of Whack a Mole. It just never ends. And so this is where Incogni comes in. They do all the hard work for you. They reach out to these companies and they will get you legally removed from these lists. And I know because last time they sponsored my show, I signed up and I didn't take the free option that they offered me, wanted to pay for it. I wanted to see if you get value for money. And they removed me from 79 data broker lists. And so I've stayed on, I've stayed a subscriber and I have seen a massive decrease in the number of emails and phone calls I've been getting. So it's a great service. I recommend you check it out. If you're sick of this like I was, please head over to incogni.com Peter and sign up. If you use the code, Peter, you will get a lovely 60% discount. So that's incogni.com Peter. So would you say where Bitcoin is a censorship resistant money, this is going to be a censorship resistant AI?

[Lieberman]

That's our goal.

[Peter McCormack]

Okay?

[Lieberman]

But that is the only goal we actually even started thinking about.

[Peter McCormack]

So let's talk about one of the fears. Let's talk about Mythos. Mythos or Mythos.

[Lieberman]

I would say Mythos.

[Peter McCormack]

So again, I don't know if they said this as marketing, but they feared releasing it because they thought it was too dangerous to be in the hands of the peasants like us. Because of the ability to find these zero day hacks like that, these zero day exploits like that, there's a real fear that we may reach a moment where AI can just break into any system. Anything we've ever said, published, looked at on the Internet becomes public. You know, that just everything is released. So they feared it. They said it's too dangerous to release right now. But at the same time, didn't we get hack into it by guessing like a thing? But so eventually something like Mythos will escape. It will be there for the public to use. In a decentralized system, there are no control. So what stops a yours becoming like Mythos are more powerful? What stops it becoming having malign intentions?

[Lieberman]

I would say put in the perspective a different picture. In perspective. Currently AI is less available for people who actually do open source code. You see what I'm saying?

[Peter McCormack]

Yes.

[Lieberman]

We used to spend our time. Now we need to spend not only our time, but couple thousand dollars a month on tokens to be up time with what for profit software can do for engineers, for profit companies can do. So the underlying layer of the Internet ecosystem, FreeBSD Linux protocols this and that. The claim for Mythos is that look, there are mistakes and the systems can find those vulnerabilities. So we will withhold access to this in order to not end up with bad actors to use this against humanity. But with slowly development of systems which are available for almost free towards the open source developers, they would use it as well to fix all of the bugs before they will be even exploited.

[Peter McCormack]

I'm going beyond the bugs. That's the bug level. But the one of the big fears with AI is when it starts to act dangerously malign intentions. Now nuclear is a great technology. We all agree nuclear is great technology for creating power.

[Lieberman]

Used as a weapon.

[Peter McCormack]

Yeah. Used as a weapon. Yeah. I mean we can make arguments has made us safer or more dangerous, who knows. But like just for power it's a great technology, but we don't want every person in the world to start building nuclear reactors in their home. We want that just in certain places. If we have a fully censorship, censorship resistant open source AI, are we essentially releasing the future super powerful super intelligence to everyone? And do we even know the consequence?

[Lieberman]

Are we releasing or are we creating the competition of billion people using the

[Peter McCormack]

same AI and it be the same thing?

[Daria]

So you're right that if we look at all potential scenarios. There is scenario like that and it has a probability. What we say there are more likely scenarios, like bad scenarios, which don't have just like 1% probability, which have 20% probability, which we should also care about in the if we really create superior intelligence, it is likely we won't be able to stop it. We all should admit the killer switch is unlikely. We will be able to invent it.

[Peter McCormack]

But if the state fears it, there are bottleneck constraints around three or four companies.

[Lieberman]

Three or four?

[Peter McCormack]

Well, yes.

[Lieberman]

And that's when we get to the cold war of multi parties actually threatening to each other with the nuclear power.

[Peter McCormack]

Sorry, but what I'm saying, there are the bottlenecks, there's the energy infrastructure, there's the data, there's all the different parts

[Lieberman]

that Chinese will have more data centers and more energy.

[Peter McCormack]

So you're now bringing to the table, is it AI 272020 27. But it's a fair discussion.

[Daria]

So the question is that we don't see a scenario when any of AI labs will stop development.

[Peter McCormack]

No, I agree.

[Daria]

Because they won't be able to stop, because others will not. So they continue both domestically and internationally. Moreover, we don't expect that some of them will be able to keep access to these models from anyone else. Still, someone in these companies will have access. Why the hack happened because eventually you don't have this in sandbox which no one have access to. Moreover, when they do this, when they limit the access before the superior tensions, they, they, they, they, they, they do it right now. They actually don't allow everyone to be more protected for the the future AI which will come. Actually the gradual deployment of this model, rollout, rollout of these models actually helps us to be more and more prepared. Because the dangerous thing is where this jump of the capabilities will be too extreme. Too extreme. But if it's gradual, then this superior intelligence will not be fighting with the humans with spears, but it will fighting with the humans with one version of AI earlier.

[Lieberman]

What's interesting also is the power in a decentralized network. How would superintelligence aggregate this amount of resources coins to pay for its inference? Because it's millions of actors who need to provide these resources to this superintelligence. For the superintelligence to be that powerful, so it can actually make a significant harm inside of a corporation, all of the resources they have, they can click one button and all of the resources will go towards a specific model. So in a decentralized network, the highest the chances are Much higher that there will be multiple actors with a lot of resources inside the same network openly competing with each other. Because openly because it's blockchain, everything is open there.

[Daria]

Also I trust much more to open source communities that they will be careful and they, they will when they train they will actually do do the due diligence. Like for example right now most of the AI safety people, they struggle because they don't have enough enough compute.

[Peter McCormack]

They seem to be quitting.

[Daria]

So no they they, I mean like independent resources.

[Lieberman]

They seem to be all quitting. Exactly. Because inside there doesn't have enough resources but outside they have even less resources.

[Daria]

So we have friends who give grants to different research groups. And what they found out that was the last year. The only reason the shift changed shift was there that 90% of the grant request is for compute. Because AI researchers also need the compute to test the hypothesis to actually make safer. Make it safer.

[Peter McCormack]

Does does if you're correct and you're successful, does your decentralized AI network weaken the nation state?

[Lieberman]

That's what, that's what we expect. Maybe even hope but not weaken in the sense like making them weak restructuring.

[Peter McCormack]

Sure. But in that scenario does the country who operates on, on centralized LLMs, do they have, do they have more power over the country that is now decentralized?

[Daria]

So we, we see that the opposite. So what?

[Peter McCormack]

Well I'm not sure you do. You gave me a look.

[Lieberman]

No, no, no, no, no. So I wanted to restructure the, the, the position here. Currently Chinese doesn't have as many resources as the US have so it will take them another several years before they will even be able to get to the level where US stands. Unless the technology actually developing towards increasing efficiency significant like 10 times every year the token is going to be cheaper. 10 times every year the new chip's going to be more productive than 10 times more productive every year. Unless we see this and in decentralized network we see that on an example of Bitcoin, the building of this infrastructure can be much faster specifically because it's everywhere and nowhere and specifically because everyone is participating in their own interest. So can the Chinese have the most of the resources? Probably maybe, but I personally doubt that they will out. If we will be able to out compete us, we will be able to out compete Chinese for sure.

[Daria]

And that's what we also see in terms of national states that yes there are these two major players but the rest of the national states they already right now concerned and not concerned that they will not have the power over their people. But they concerned that actually they will need to obey to either yes or China positions because they will need to have access to. And from this perspective, what we found out that actually the rest of the

[Lieberman]

world, the rest of the 200 states

[Daria]

are super supportive of decentralization.

[Peter McCormack]

Well, of course, because they don't have access.

[Lieberman]

Exactly.

[Daria]

Because actually this, they have the power from anywhere.

[Lieberman]

They have all the resources to build the most powerful decentralized network if it's their only alternative. And this is in a sense their only alternative if they want to have access to the most powerful network which will be decentralized because each of them cannot have their own most powerful system because they're much smaller in terms of economical power than Chinese or us, Then together combined they will be able to build the Internet of AI.

[Peter McCormack]

Well, I also, look, I am, I'm of the position that properly decentralized, you are stronger than a centralized network. And I think of that in the world of the nation state and bitcoin. So let me give you an example. At the moment, the UK where I live is a shithole. It's collapsing, it is turning third world, is going very Marxist communist. The future for the UK is, is not great at the moment. But we can't out China China because we still want to give off the air of freedom, the air of democracy, that we are free nation, blah blah, blah. So we're like, we're doing crappy China at the moment and so China beats us. I've always felt like if we did the opposite of China, we're completely decentralized, decentralized, the money, decentralized power. You are stronger because you have a much more strong network is because the weakness comes from a few people, a few choke points. If you distribute, distribute the choke points, you become a lot stronger.

[Lieberman]

If you build economy which consumes 6 billion people, not the billion of the Chinese, you're stronger, you're stronger. I agree with this completely. If you increase the speed of, of economical relationship of the market by doing this decentralized and on chain and all digital and no gatekeeping, you actually build a stronger economy. If you can build a stronger economy than the China or the us you're winning.

[Peter McCormack]

So the journey from where we are now to where we are in 10 years, I feel like this is the scary period because we don't know what's going to happen.

[Daria]

And unfortunately there are also old current powers are like against pushing, pushing hard

[Lieberman]

and fast against what we are saying. Like they are pushing in different world, completely different world. Even if they don't do it on purpose, it's just the way they do. They do stuff today.

[Peter McCormack]

But they are a small group of people. They're completely compared to the very large group of people who have skin in the game.

[Lieberman]

On the downside, what's interesting, even in the United States, we should understand that there's like not more than a dozen corporations which are weaning from what's happening right now. And the rest hundreds if not thousands of billion dollar companies in the United States. They're all losing from these 10 companies winning. And they're all concerned. So for even for them, even for the rest of the United States, having a decentralized, completely equally accessible public AI network is a win.

[Peter McCormack]

I again want to be careful with my wording, but this also feels like a war footing. So what I mean by war footing is that I did an interview this week with Andrew Wilson. He's a conservative guy. He refers himself as a Christian populist. He's very concerned at birth rates. The, I think the, I think the birth rate we require for stability is like 2.3 children. But really he thinks we need to be getting back to three or four. He's very concerned by that. He's concerned by the propaganda that's used to kind of subvert the family unit. And he said that we essentially in a war footing, trying to reverse this trend. We want people to be able to own a home, get a job and have children. We want that. That's really important. And I agree with him because what happens to humanity? We're not replacing our old people with new young people. But he refers to it as a war footing. This is so important. He's on a like, consider it like a war. Again, being careful with the term war, but at the same time the future of humanity is being challenged by AI because there are just a variety of people who have a variety of different kind of thesis on what is going to happen. Like, will the AI kill us all? Will we end up in the Matrix? Will it be ready, player one? Will it be? Who fucking knows? Like some people say, oh, you're being crazy. None of that shit's going to happen. Other people are like, oh, I'm not going to be an AI safety researcher anymore because we're all going to be dead in three years. Nobody actually knows. We just have ideas, right? But if the majority of people are going to see their lives hollowed out and their experience of life become harder because increasingly labor is going to AI and not being replaced, and we live in an Inflationary system whereby they're just running to keep up. We are essentially in a war footing to try and as all the peasants come together to fight the powers.

[Daria]

That was exactly why we called the protocol.

[Peter McCormack]

Yes.

[Daria]

Which means the race. Race that we actually don't have much time, you know, to, to. To actually historically create arms race, in a sense, to create the alternative. We maybe have like one year, two years, and then it will be too late.

[Peter McCormack]

How do you convince people, though? This is the challenge.

[Lieberman]

So I think that's why we're here.

[Peter McCormack]

No, because decentralized technologies always are a challenge on the user experience. If you look at Bitcoin, using Bitcoin, when you're first using it, compared to like pounds and dollars, pounds and pence, dollars, cents, they're very easy to use. It's like, oh, there's a hundred cents in a dollar.

[Lieberman]

Bitcoin was impossible to use.

[Peter McCormack]

Yeah, well, it's not impossible. No, no, no. It's actually, I think it's very easy to use, but there's a friction the first time you use it because it's different from using the dollars. Plus, also, if you make a mistake, you send it to the wrong address, you can lose it. And also you send this long string of characters and you know, you have to worry about being hacked and stolen. There's all these things that is new. And the original wallets were. First you had the command line interfaces, which were very tough to use. And then we got the simple. It's just a challenge. Can what you've built look like. For example, I've played with a couple of the privacy focus AI tools and they're just not as helpful and easy to use as Claude. And so how are you able to build something that feels as good and looks as good?

[Lieberman]

That's why we started not from a chat or a tool, but from API.

[Peter McCormack]

Okay.

[Lieberman]

That's why we actually honestly not building anything but the API, because as soon as you build API, you're actually focusing on others developing something on top of your API. And the whole point of the Bitcoin tokenomics, like Bitcoin like tokenomics, which we decided to utilize in our network, is in ability to subsidize the growth of this network from a perspective of the coins being mined and issued, and therefore the miners being rewarded with the coins issued, not with the coins paid. There's a balance between coins issued and paid. So at every particular moment, coins issued and paid together combined need to be higher than the cost of mining, than the cost of providing the hardware hosting the models somewhere.

[Peter McCormack]

Hold on, just. Sorry, take me back a step. Where does the mining, what role does the mining play in your infrastructure?

[Lieberman]

So in Iris infra, imagine that you

[Daria]

have a GPU server which can run

[Lieberman]

a model, let's say Deepset or Kimi.

[Daria]

Then from your perspective, you don't need to think about the developers, you don't need to think about the other sides of the market. You don't need to think about sales of renting this out. You just install the software the same way as you do with bitcoin mining and you participate in mining and you get your coins.

[Lieberman]

Now you can sell the coins or hold the coins, it doesn't matter. Depends on your risk profile. But what you do is essentially you provide the AI services without even knowing that you're providing the AI services. For a decentralized network, then people connect with API and they pay for the API requests, but they can pay much less today. Today particularly, the price is going to be tens of times lower than anything you can find on the market today. So our vision on this network winning is because of this structure, because of the financial insensitive, additional financial incentive for the miners to join the future worth of the coin, the current worth of the coin, they're getting the subsidy in a sense for providing AI inference much cheaper to the developers, much cheaper than anything else on the market. And it's API specifically so that others are building the builders who want to build better products using this instead of using, instead of using OPUS or GPT.

[Peter McCormack]

I understand.

[Daria]

Moreover, what we see right now, you know, like people, like humans, they are really like they. You see the inertia, it's really hard for them to switch from ChatGPT or Claude if they already.

[Lieberman]

You said yourself you was using GROQ and ChatGPT before you actually discovered that Claude is actually doing a great job.

[Daria]

Yeah, but what we see already right now with the network, that agents are not like that agent. As soon as, for example, some users installed OpenClaw and use anthropic model, usually after a week they see this bill with hundreds of dollars, hundreds of dollars a week. So they just ask the agent, can you find a way to be cheaper, cheaper? And the agent itself then have the single goal to find because it's open source. No one affects that single goal to find the better, the cheapest provider.

[Lieberman]

And as soon as its agent itself is looking for the cheapest provider, it will end up finding open router. And inside of the open router they will find open weight models which are already cheaper. And the providers which are already cheaper. But as soon as our network is there, they will be able to find our network API and just the success of the network is guaranteed.

[Daria]

Originally when we built it, we thought that most of the usage will come from developers who develop different apps to their users. But what we see right now, most of the usage is actually agents and it's already right now. And that's the reason why Jensen on the recent conference mentioned her claim that

[Lieberman]

this, the open claw is like an iPhone moment for the AI.

[Daria]

Because the usage of AI actually tripled through the last couple months because of

[Lieberman]

AI. It's tripled because of agentic systems, because of systems like GStack, Gachitan, OpenClaw, Hermes and others.

[Daria]

So in that's where we believe decentralized networks will play even more important role. Because imagine the system where the decision for provider of AI service is made by AI and the only goal is to make like better service for user and cheaper for user for user.

[Lieberman]

Better and cheaper service for user.

[Daria]

So they will likely more and more lean towards. Lean towards decentralized.

[Peter McCormack]

Okay.

[Daria]

Moreover, it will be easier for them to pay because of the crypto. It would be easier for them to pay. It's natively for them to pay in digital currencies rather than the regular using the bank.

[Peter McCormack]

So let me ask you, do you consider yourselves as businessmen or do you consider yourselves as activists?

[Lieberman]

More of activists.

[Daria]

So we already built companies before in sold company. Our company In 2016, a computer vision company was acquired by Snapchat for 60 million.

[Lieberman]

This was the first time we were ever employed by anyone Snap. But before that it all started with activism. So 2001 it was all about activism for us. Building open source software, building decentralized network. And then we realized that there is another way. And this another way is venture capital growing. Back in Moscow, we didn't have actual access, but we just was reading and observing from the other side of the world what is going on in Silicon Valley. And for us was like this is it, this is it that Google is doing good and changing the world. All of those corporations, all of those no corporations, but all of these startups companies, they raise capital and they build systems which then can change the world for better. This was.

[Peter McCormack]

And make some money.

[Lieberman]

And make some money. So we started focusing on, on startups. And with every single startup we were looking at an angle through which we can continue our activism. Like we, we started developing an online game because we saw that we will be able to promote freedom and show different coordination in. In the massive multiplayer online game. The business end up to be much harder than what we expected. So it failed. But then after that we started focusing more on computer vision as something which was sort of natural. Like you develop computer graphics and you move to the computer vision. And that's where the major, major financial success came to us. But for us, it was always about the mission. And the mission was always about increasing the, the, the level of freedom to the extent when we will go past the national states.

[Peter McCormack]

So, so as an activist, this. You're really building a company or sorry, not really a company because there's no company. You're building a product because of what you fear is going to happen in the world.

[Lieberman]

This particular. Like we've been building this businesses to make money and then spend money on the activism. And then we realized three years ago that's it. Like it's all.

[Daria]

There's no.

[Lieberman]

There is no time for us to make money for the activism. It's either right now we will shift all of our focus on building decentralized AI or we're done with all.

[Peter McCormack]

This is existential.

[Lieberman]

This is existential for our vision of what we wanted to build in the

[Peter McCormack]

world or the world you want people

[Lieberman]

to live in the world we will be. Exactly.

[Peter McCormack]

Yeah. So what is your kind of. What is your, what is your message? Like the mission that you want people to really hear? Like if somebody's listening to this now and they're like, yeah, I'm using Claude, whatever. What do you really want to learn? What message do you want?

[Lieberman]

Listen, so currently, when we pay for all of the AIs, we pay for the corporations to build the infrastructure and they will build it using our money to then replace us. That's the funny irony behind what is going on. We all understand they will replace us and yet we keep paying them for them to replace us. For them to build this massive infrastructure, spend billions of dollars. And they can only raise this money because investors expect that we will be paying them subscription or per talking or per API request payments. And that's how they will get the revenue for them to then replace us.

[Peter McCormack]

They're like heroin dealers.

[Lieberman]

Exactly. Exactly.

[Peter McCormack]

Here's a free one instead.

[Lieberman]

Instead, we can spend tiny resources right now to help the decentralized alternative to be built. And look at how it's developing. If anyone is just switching to use our network A, it's going to be much cheaper. Yes, it's maybe not going to be as great as Opus because it's Kimi or Deep Seq. And at the same Time the latest Kimi or latest Deep Seq are quite good. Like they will be in top 10 top 20 best models in the world of the frontier models. Not just like any, but like comparable to frontier models. So the service is going to be that cheap that it's going to be. It will not matter as much for this to be not as good as OPAs, let's say specifically for agents. And we can see this on the open router data more and more in the top 10 used models through the open router. It's open source models because cheaper means better when we're playing the agent game because agents are reasoning and agents are spending millions and billions of tokens to achieve the results. And you need the result. You don't need like the precision in the same sense when you writing a text through.

[Daria]

So instead of paying to these corporations to build infrastructure which they will eventually sell to us, we can build infrastructure together and have the co ownership by having the tokens of the system so that we all benefit from the infrastructure which we together built.

[Peter McCormack]

Yeah. So how does this work on the next layer with the. So the power is the big bottleneck right now. People talk about power, access to power. It's like gold dust at the moment.

[Lieberman]

You mean electricity?

[Peter McCormack]

Yeah, electricity. So there's. There's companies who've built massive infrastructure, massive data centers to. To power this. How in this world is does that. Are they still a supplier? They're still a necessary supplier. They've they've just got a different customer.

[Lieberman]

Power is not problem. Power is only problem in United States

[Daria]

because when it's concentrated and when it's

[Lieberman]

concentrated in a specific point data centers which provide inference does not need to be in one place because the model itself when you send your request, not that it's actually being like. Not as it's using thousands of GPUs, it's only using eight GPUs in one specific server. The model itself fits in a memory of shared memory of 8 GPUs and that's where it is. So if a server with eight GPUs standing in Africa and another one standing in Algeria and another one is Chile, like cross the world.

[Peter McCormack]

But sure. But what happens to these big data centers? Can they just switch and plug into this?

[Daria]

Actually is that what's happening? Partially. Like for example in, in our network first we saw professional miners, people who mined Bitcoin before Ethereum, before they know what proof of work means, they jumped in and joined. But then we also saw several new clouds. The smaller Data centers, not AWS, but smaller data centers with 20,000 GPUs and they wanted to just try it out. So they connected some of their spare GPUs to the network and then in a month they saw that they actually earn much more than from the GPUs they were renting out. And they added more and then they added more. So from this perspective, this principle. So for example, United States today consume around like 1.2 terawatts while all the data centers consume only less than 20 gigawatts. Less than 20 gigawatts.

[Lieberman]

And then AI itself is probably around five in the United States, five gigs.

[Daria]

It's actually tiny parts. What problem right now that they are trying to build this massive data centers just concentrated just in one place. And that's where they get all of these troubles with power. Actually people around the world have much more available power generation capacity than the

[Lieberman]

Pacific Data Center Corporation. What's interesting is some might say yeah, but they are everywhere in the world. Therefore the ping is going to be worse. But the trick is the delay, the latency, latency mostly is coming from the GPU work, not the network. So the network is going to be like what the bat ping is going to be 300 millisecond while the model itself is actually like the time to first token is like over one or over one second.

[Peter McCormack]

Sure. But like if I wanted say myself, I'm like okay, I love this, I want to run this, this is cool. Could I go to a data center say look, I just want to partition some power for me.

[Lieberman]

That's what, yes, that's what people do right now the majority of the GPUs in our network right now is this people who believe that this need to happen.

[Peter McCormack]

What's a bit like Amazon Web Services, right?

[Lieberman]

They could be like there are new clouds renting out GPUs and people just come to them like together AI G, Core, Nebus, Hyperfusion and others who providing

[Peter McCormack]

GPUs for companies and they hyperscalers providing GPUs.

[Lieberman]

They're not like there are hyperscalers like Amazon providing GPUs as well. It's just more expensive with them. They're like smaller, mid size, small mid sized data centers providing GPUs to companies, to startups. And they are very efficient in the sense they need to compete with the hyperscalers and for that they need to be cheaper and more efficient. And people right now, majority of the almost 4,000 GPUs in our network right now comes from people Renting out from the Neo clouds and then installing nodes on these servers.

[Peter McCormack]

What happens if you're not successful? Like what if the safest thing for humanity is not what the market chooses?

[Daria]

Unfortunately, what what we expect in that case that models will be.

[Lieberman]

Models will divide in two tiers, the closed tier and the public tier. What we have right now is public tier and the closed tiers are only going to be available for governments and selected corporations like 10, 20, whatever, whoever is complying with some agenda behind what's going on politically and economically within this corporation.

[Peter McCormack]

How everything works all the time.

[Lieberman]

How everything works except at the extreme.

[Daria]

This time the power will be power reserve systems.

[Lieberman]

I'll give you an example. If the model will be capable when not if, when the new opus, like two, three versions from now will be capable of rewriting a software by just saying like hey I like this app, create me a copy of that one but completely new from scratch so that no one can actually even say that the code is the same. Like no similar line of code. So it's completely different app but doing exactly the same. Then a corporation like Google can easily say, oh there it is. We have a statistic of the top thousand applications in Google Store which are mostly used by people. We don't need the rest of the hundred million applications over there. We just close the Google Store whatsoever. And we then give subscription Access, let's say $200 a month to all of the thousand top applications which most of the people use, completely rewritten by their own algorithm. And now they can charge 30% of the third party developers but 100% for themselves.

[Peter McCormack]

The technical answer, what happens if this,

[Lieberman]

what happens if this is happening is we are moving towards economical dependency. Yeah, like specifically of younger experts, professionals.

[Daria]

We should expect the government will have more control. The government, you already see, hear it from people like Eric Schmidt, that likely printing labs will be nationalized like that or like semi nationalized through China.

[Lieberman]

Everything in China is already semi nationalized.

[Daria]

Yes, with Chinese model we expect the same. So they right now open. So they right now open source access to the models so that you can actually download them and install on your own server. We believe that in a year from now it won't be like that maximum

[Lieberman]

two years, but likely in a year

[Daria]

from now because they use open source just to catch up to get to the frontier state. But then they likely will close. We already see it. So that they actually delay the public launch of the newer model, open source launch of the newer models and so you need to use their service which is much more expensive if you want more controlled and more controlled like for sure. Like sensors and control.

[Lieberman]

More sensors and more control.

[Daria]

So that's what we expect.

[Lieberman]

Less freedom. Like, clearly less freedom. Like everything is going to be controlled for the sake of our own safety.

[Daria]

Yes, it will be. Always for the sake of our. Always.

[Lieberman]

For the sakes. For the safety of our kids. Always say safety of our kids always. And therefore indisputable. And that's only the question of two, three years from now.

[Peter McCormack]

You think within two, three years we're going to descend into an authoritarian, dystopian future.

[Daria]

Our hope that decentralization will know what

[Peter McCormack]

your hope is, what your fear is,

[Lieberman]

what our fear is. Because look at what's happening with war everywhere. Therefore more safety, less access. We will see some drone attacks.

[Peter McCormack]

Why two to three years? That seems.

[Lieberman]

Because look at the speed of development currently. Look at how far we get from what was three years ago

[Daria]

in three years.

[Lieberman]

AI models right now. AI models right now helping to write new versions of AI models. It's already happening right now.

[Peter McCormack]

That's the recursive loop.

[Lieberman]

That's the recursive loop. Engineers are not mostly not writing their own code themselves. They use AI to write the code and then they review and then give another task, another task. And they launch multiple agents at the same time doing different parts of the code and then reviewing and combining this. All this is efficient and this is empowering you like you can create more, much more than we could have imagined before. So the speed at which this technologist will be developing is going to increase.

[Peter McCormack]

So could you be too late?

[Lieberman]

We may be already late.

[Daria]

Some days we wake up with this feeling that maybe it's too late. You know, like every time, every time there is some announcement that of another

[Lieberman]

$100 billion invested in the data center there and there we're like.

[Daria]

So for example, recently in the last months we were reach out by some of the miners in the network were asking, do you know where we can get GPUs? Because there are no available GPUs in the market. So they wanted to rent out to. To. To actually some people who wanted to

[Lieberman]

put like millions of dollars into mining. Gonka, they cannot find GPUs right now. It's hard to find GPUs right now for the. Especially for the short term run.

[Peter McCormack]

Do you need the Jensen GPUs or can you use other ones?

[Daria]

Currently we use. That's our hope. But it's hope, you know, like our hope is that actually because we already have really great people in the community who previously were developing ASICs for Bitcoin that they actually already right now really with designs of new type of hardware which will accelerate AI like 10 times at least next year. But they need to be in time. And we hope also that they won't be just acquired by Nvidia as GROK

[Lieberman]

was or as represent is being merged submerged into OpenAI and stuff like this.

[Daria]

So for that we need to be sure that they actually got this incentives when they understand that they don't need to be just sold. So a corporation that they can just produce the hardware, connect to the network, get the tokens and then continue this loop.

[Lieberman]

So what we have what's interesting, if people just. If the price of the coin will just go up, then for the people who own GPU right now and who rent them out to some other companies and services, they will, they will be able to see that they can make

[Peter McCormack]

much more money financial incentive.

[Lieberman]

Financial incentive. They will be able to make much more money from installing the node of the decentralized network. And therefore actually honestly even then providing this, it's not just eating this hardware as Bitcoin would have done, it's providing the inference cheaper and then you make money out of the coin price going up.

[Peter McCormack]

You said you wake up some days and you're like. And you've also were part of a group that kind of thought about different scenarios. What is the darkest scenario that you almost don't want to say out loud?

[Daria]

Darkest scenario is that actually. First, if you cannot produce anything which other people would want to consume. And it's a big question of meaning in the life and from this prism, the darkest part, if we won't be able to coordinate, if we won't be able to change the course, then there won't be humanity anymore. It's just. Can it be something more darker than that same.

[Lieberman]

I would say that the darkest will be the war. The actual real war.

[Peter McCormack]

What is the war? Is it a war between nations or is it a war against the machines?

[Lieberman]

Not against the machines. I think that we're not even get to the point of the war against the machine before we'll start the war against the corporations which are stealing all of the jobs and the nations which are actually benefit most of this development.

[Peter McCormack]

You see a war between the poor

[Lieberman]

and rich, the war between those who have access and not.

[Peter McCormack]

Well, that is a war that's happening right now. It's a war of words.

[Lieberman]

It is happening right now. That's why I'm saying like it is happening right now.

[Peter McCormack]

It is happening right now in the UK with the fastest growing parties, the Green Party, led by a guy called Zach Polanski, who is a socialist, therefore communist.

[Lieberman]

Leaning towards the ideas of the past is not going to help us.

[Peter McCormack]

Well, certainly ideas have never worked, especially

[Lieberman]

the idea which never worked. Like look at what's happening with Russia right now. Did the past help it? Not even now. Like, they are looking in the past and they're hypnotized by the failure of the future. This is a trauma, national trauma, which leads them to starting the war with neighboring nation. And millions of people are dead already in the course of like four years.

[Peter McCormack]

Millions, Millions. I read this morning that Putin's popularity is falling fast. Can you talk about this?

[Lieberman]

Unfortunately, we cannot get back to Russia.

[Peter McCormack]

That's why we can't go back.

[Lieberman]

We can't go back.

[Peter McCormack]

You enemies of the state.

[Lieberman]

We already said too much.

[Peter McCormack]

You said too much.

[Lieberman]

Unfortunately, maybe not too fast, but as always, this might collapse and will collapse unexpectedly. And that's what we hope. But at the same time, the kgb, the fsb, whatever they call themselves right now, and Cava there they used to call themselves 100 years ago. They are very adaptive and survivable structure of psychopath.

[Daria]

This is the issue. We experience that all the signals are visible quite early, but people just don't want to react. For example, in 2012 there were a huge attack within social media on every opposition mined in in Russia. And we made an investigation and found out that it was made by the network of hundreds of thousands of bots which were all created in 2008. It was accounts registered in 2008. Then they had connections to real people. They, they had, they were having conversations.

[Lieberman]

KGB was infiltrating Facebook earlier than anyone could have imagined.

[Peter McCormack]

I think I heard Ron Rogan that 19 of the 20 biggest Christian Facebook groups were Russian bots.

[Daria]

Yeah. So at that moment we approached some of our friends working at Facebook, working at

[Lieberman]

high enough in the chain and

[Daria]

their reaction was, Russia is not as important market for us, so we don't

[Lieberman]

have to deal with this now in price.

[Daria]

It was just like what?

[Lieberman]

Like you don't understand.

[Daria]

Eventually it will affect everyone and it won't be and we won't be able to stop it.

[Lieberman]

Part of the reason this war is still going on is because the west was delusional about what was going for real. Like the largest army in the second largest army in the world and then obliterated by the smaller army of Ukraine.

[Daria]

So from this perspective, what we see that signals of this bad Scenarios, they're quite visible early, but you need to pay attention. The same right now happening with AI. The signals out there. You don't need to be people just like the person on the earth to

[Lieberman]

see that the Mythos situation is just the marketing. Yes, they probably use this as marketing, but this is also a signal. And they look at the reaction. Is everyone okay with them having two different tiers of access, the special access.

[Peter McCormack]

And because they've given access to the companies the Apple selected companies select. So they for example, this is propaganda.

[Daria]

Yes, they, for example, concentration of power. They, they, they mentioned that this is for security because they need to secure major banks and then they gave it to JP Morgan Chase.

[Lieberman]

But what happened with what with other banks?

[Daria]

With banks or like with European banks. So, so if you really concerned about protecting people around the world, the rollout will be different. But here the rollout is the companies. Who pays you more? Is it concentration of power? These models have this recursive nature that if you give someone the latest model and you won't give it to everyone else, there's someone getting fair advantage in the market. And all of us are ready to pay. So it's not about money. They just gate who can get access.

[Lieberman]

Now they have more money to pay back and we're all losing. Now we have less money, so we cannot pay. And this is a recursive structure through which pretty soon, really soon, we will find out that this chunk of US corporations are paying most of the money. Imagine what happened with Nvidia for example. Nvidia is selling their GPUs. Almost like 90% of Nvidia GPU is going to like five, six US corporations. Not that anyone in the world would be ready to pay. Everyone in the world will be ready to pay. It's just like the decision made back in 2023 to concentrate in US. It's a political decision and a power decision rather than economical decision.

[Daria]

Moreover, the entire development of Nvidia happened because all of us Gamers used our GPUs and made advancements could which right now used for AI mostly were used by game developers around the world actually contributed to that development. But now it's gated. Now we use it against us.

[Lieberman]

It sounds too dark. Scary and too dark. But honestly at the same time we are very optimistic because the technology itself eve rightly done. And I'm not saying like safety or something distributed equally, accessible, permissionless, permissionless, immutable. All of these words which we learned the hard way, we created them through the last 16 years. Of the blockchain development,

[Peter McCormack]

bitcoin development, bitcoin development.

[Lieberman]

I agree with you. Honestly, I agree with you. It's all about the bitcoin. All of this diva, all of this principles, they showed efficiency, they showed market efficiency and they showed the, the public benefit efficiency. If we will be able to coordinate right now similar but faster to what we did with bitcoin, and we really hope that it is possible just because imagine, yes, Bitcoin is 16 years. Yes, the major development only happened throughout the last 10 years. But there are people who truly believe and there are millions of them and there are millionaires because of the bitcoin.

[Peter McCormack]

And you need that community to come to this.

[Lieberman]

We need that community to come to this. Well, and we're not saying like, oh, we're building next bitcoin, so go buy. Like fuck this shit.

[Peter McCormack]

Oh, this is a mission.

[Lieberman]

Whereas what we're saying is we need to repeat the principle of bitcoin in order to build a completely different infrastructure. Not the financial tool, but the AI tool. Repeat the history to build not a competitor to bitcoin, but to build a competitor to OpenAI, Anthropic and Gemini.

[Peter McCormack]

I think you need to really think about your 10% of coins and how that changes how people perceive what you're doing. You don't need to answer that now.

[Lieberman]

But what we also did, we took another 10% and we actually issued it. But we put it on the smart contract of a DAO controlled by all of the miners.

[Peter McCormack]

But you've already sold a company successfully 60 million. I'm just saying, I think it changes a lot of the incentives and the vision and the picture if that, if this doesn't benefit you financially on the upside. Do you understand the like, if you want to do this, you. I mean you can't hide who you are now. But. But you have to be the satoshis of AI. And if you have to be the satoshis of AI, it has to be mission without financial upside.

[Lieberman]

There's only one satoshi and it's satoshi. We cannot be the satoshi.

[Peter McCormack]

But you can be. You know, seriously, I mean if you're right, the just for me, that's the message. But like it's not a question to answer now. This is a philosophical question.

[Lieberman]

It is the resource. This resource we honestly believe we will be able to use towards the vision which we want to achieve.

[Peter McCormack]

You can sell it's in this world. We can sell hope and we can sell fear that that is what it is. We sell hope and we sell Fair politicians sell hope and they sell fear. Companies sell hope and they sell fear.

[Lieberman]

So the fear is already there. We're not selling you the fear.

[Peter McCormack]

The fear is there. I know. I look, I see on multiple levels the fear of what's happened to my job, what's going to happen to my kids, and fear about what this technology is going to do. Is it going to kill me? I know the fear is already there. Like I fear it myself. I fear. Do you know what I really fear? I fear I'm wasting my time working over the. The next few years of not living the human experience. Because what if something bad happens? I wasted my time on airplanes, in hotels, when I could have been with my kids.

[Daria]

Exactly.

[Lieberman]

This was the reason of my depression few years ago as it started and then stopped with this new fear, with the new fear of what's going to happen with AI and it's going to happen fast. And then I just basically woke up like, oh shit. We need to focus on this for as much as possible to make an attempt to build an alternative. I'm not saying obviously this is almost as if a David and Galiav situation. So we are trying to challenge not just a handful of US Corporations, but the political decision of the previous and current administration. Both it was Biden's administration political decision to hold all of the GPUs in the United States. And listen to what Jensen is saying. He's like, cannot you see this is only proper right decision to make this where like we have to concentrate it here. This is the way it must be done. But there is another. I can see how Sam or Jensen sees this as a positive. Honestly, I'm not trying to portray them as evil people. I honestly believe they want the good for the humanity. It's just a question of they have not seen the different paths. OpenAI started as a nonprofit and converted into for profit. Not because Sam was greedy. I truly believe it's not true. The reason they converted a for profit is Sam's understanding and vision that if he wants to be to build the AI which will save the world, not the Google or Microsoft or Amazon, he needs to get resources, enormous resources to build data centers with millions of GPUs to compete with those corporations. He just didn't get to the, to the, to the same idea we did, that we can build it all as a bitcoin. And maybe he was right. Maybe what we're doing right now would have been impossible without what already happened with OpenAI and Anthropic and Gemini. Like without them actually doing this, we wouldn't be able to get to the point when we can offer a different solution.

[Daria]

So

[Lieberman]

the positive view on this is that it is a natural way of things. It couldn't have happened differently and this is happening right now. But we've seen horrible tragedies in the past, like the world wars of the past and financial crisises like a Great Depression, all this shit. So not that all of the scenarios will be positive as much as they are natural. Currently we are racing towards the wall towards a really bad outcome. Racing. And there is a small chance that we will divert this by combining our efforts into building a decentralized infrastructure. I hope it's optimistic enough.

[Peter McCormack]

Gonker, I wish you guys the best. I'm gonna go and take a look at Gonka now. And I appreciate your time today.

[Daria]

Thank you for having us.

[Peter McCormack]

Thank you to everyone for listening. We'll see you soon.