Hi everybody, this is Charles Hoskinson broadcasting live from warm, sunny Colorado. Today is February 4th, and I want to make a quick video to talk about something that has never happened before in the history of the U.S. government: an outside group of people who aren’t elected but appointed coming in to audit everything.
What provoked this video was a tweet I saw that brings up a valid point. I posted something about what’s going on right now, commenting on something AOC said, and someone asked, "Would you consider Musk to be an unelected bureaucrat?" I responded that this is a talking point from the DNC, which is essentially saying that Musk is like a bureaucrat, coming in with all this power.
The point I was making is that Musk is most likely to outsource a huge amount of his decision-making to Grok, which will be a superintelligence this year. By extension, he’s one of the first augmented superhumans. Having unlimited money and computers really does make a difference these days. He has already raised $6 billion for XAI, and its valuation is over $60 billion, which is actually more than he paid for Twitter. He used Twitter to launch XAI, so Twitter is now profitable for him.
Musk has an enormous amount of compute power at his disposal. Llama 4 is coming out in about a month or two, and it will be significantly more capable than Llama 3.3, which is already as capable as the prior frontier model, ChatGPT-4. When you look at Llama, one of the top things there is DeepSeek R1, which you’ve all been hearing about. It’s a thinking model, and Llama 4 has something called a dual-former architecture, allowing it to do both slow and fast thinking. It’s like a combination of Llama 3 with GPT.
DeepSeek has 7.8 million polls on AMA, updated 13 days ago, which is pretty exponential. The next Llama is coming out, and the thing is, Musk gets all of that for free. They can backward engineer it. In fact, there are four independent engineering teams at Meta right now that are tearing apart DeepSeek to understand how it was built. They will use those lessons to augment Llama 4. There are already many people looking for inspiration from the design of DeepSeek for their own products.
I tweeted earlier today that O3 Mini is two orders of magnitude cheaper to run than O1, despite being as capable. What this means is that Musk wants to be the smartest guy around, and he’s going to take all of his resources and run them into a purpose-built version of Grok. The Doge team is getting read access to everything in the government. For example, the Treasury system is a great case. The propagandists are saying, "Oh, I guess Doge is going to approve all of the treasury payments now." No, it’s very different. Read access means they can see everything.
They’re going to feed those billions of transactions, which no human being can look at because there are too many, into these super-capable AI systems. They will build templates to label certain groups of things and find anomalies, like the hundred million dollars for condoms to Hamas, for example. All of that goes out into the public discourse. What this effectively is, is radical transparent government.
Think about it: when you look at where U.S. aid is going over the next 90 days during their moratorium, we are getting radical transparency for the first time in our lives. A superintelligence, evolving at an exponential rate, is grafting itself onto the transaction layer and data layer of the U.S. government. That superintelligence is having a conversation with us about what it’s finding and what it classifies as waste or fraud.
People’s politics are radically different. Some are totally fine with LGBTQ operas being held in Colombia, while others are not. What’s nice is now we know; we didn’t before. It’s a public debate, and we all get to engage with it. This is causing some people to go a little crazy. The left always does the same thing, and the right does to a certain extent. I’m old enough to remember Rush Limbaugh and his personification of Hillary and Bill Clinton, which created derangement syndrome. Anything they touched was evil and wrong.
Similarly, the left has done the same with Donald Trump, creating Trump derangement syndrome. Now we have Musk derangement syndrome, where everything Musk touches must be evil and bad. The reality is that he possesses superhuman capabilities. He’s one of the few people alive who has the money, resources, and mandate to apply a complete AI-ification of the U.S. government. No one else has been able to do that.
This is universally a good thing for us because it will show where we’re spending our money. There will be a lot of question marks when Zelensky does an interview and says, "Hey, you were supposed to send us $175 billion, but we only got $75 billion. We don’t know where the other hundred went." He’s saying money disappeared somewhere.
Now we are on the other side of it. Every time we send money, it goes through the Treasury Department. If you get read-only access there and have a superintelligence connected to it, we actually know where all the money is going for the first time in American history. That’s pretty cool. This is not a partisan issue; it’s quite bipartisan. The response ought to be, "Okay, we think this radical transparency is useful. How do we ensure it’s used in the right way?"
I wanted to provide some elaboration on the comments. Musk continues to get more polarizing; he’s becoming more hated. It’s getting to a point where some people are calling him a Nazi. I don’t think that’s helpful at all. Now, all of his products must be bad: X must be bad, Grok must be bad, SpaceX must be bad, Tesla must be bad, simply because we hate the man. It’s a derangement syndrome.
I know the symptoms. I see the symptoms. People have Charles derangement syndrome and associate anything I do as a scam. That’s why the Ethereum Foundation can never mention Cardano in a positive light; it’s an extension of me, and therefore it must be evil. We never got a congratulations from anyone on their side for fully decentralizing it.
So, recognize when you are the victim of talking points and when there’s derangement syndrome. It’s very important to do that. Radical transparency is always a good thing, and getting a semantic understanding where a computer can comprehend what’s going on in the U.S. government is also beneficial. If you want blockchain, this is the first component for it. By having a superintelligence connected to our systems, we can get all of that data in a format where we can figure out how to graph blockchain systems to make them far more accountable to us as a community.
We’ll watch this closely. I think a lot of good will come from it. This is going to be a breakthrough year for AI and deep research. The capabilities are only going to grow remarkably. Llama will have a lot of remarkable capabilities as well. What we’ve seen with some of the open models coming to the market, especially from China, shows it’s a very competitive space right now, and capabilities are growing, especially agentic capabilities.
This is the year of agents. The next ten years will be remarkable for those capabilities. As we develop them, they will become companions to help us navigate big data and complex problems. As an American taxpayer, I care a lot about where my money is going, what they are doing with it, and what we get for what we spend. I think every person listening should care about that; it’s not a partisan issue at all.
For one, I’m very happy that there’s a group of people holding the government accountable, coming in and taking a look at all of this stuff. I understand that propaganda gets in the way, but here’s the reality: Musk has no official position in the U.S. government, which means he cannot unilaterally decide what to do. An appointed or elected official has to make that decision, whether it be the Treasury Secretary, the Commerce Secretary, the Defense Secretary, or the President of the United States.
Whether you like the president or not, he is the president, and he just got elected, which means he has a mandate from the American people. Musk is in the advisory part of the world. As an advisor, we can either choose to embrace that or reject it. If we embrace it, there’s a strong possibility that we can get blockchain into the story of government. If we reject it, there’s a strong possibility that the system will simply be co-opted and not do a great job.
I, for one, am embracing it. I believe Grok 3 will bridge the gap. It’s going to be a very powerful model and will start getting fine-tuned specifically for things that OpenAI and others don’t, because of Doge. It will get fine-tuned for government transparency, payment outflows, and all kinds of diligence and research. It’s a perfect summarization tool because of its use on Twitter. When people on X post, Grok has to summarize those posts in the news.
It’s taking a massive swarm of complex data and creating readable summaries for people. Does that remind you of anything? A massive swarm of complex data about the U.S. budget summarizing it? Eventually, this radical transparency will also extend to the lawmaking side. Every time they propose a bill or a budget, the AI agent will be able to read the entire bill instantly and let you know what it actually means. We can disagree on whether the summary is accurate or not, but at the very least, we can agree that it’s probably a useful thing.
Otherwise, the old system requires us to vote on the bill to see what’s in it. Why do we accept a system like that? We’re spending over six trillion dollars a year, and apparently, they don’t even owe us an explanation of what they’re doing. That doesn’t make any sense to me, and that’s where evil and corruption slip in.
I’m very excited about where this is all going, and that’s why I’m getting heavily involved in it. We’ll see on the other side. Thanks, everyone.
1:07:35
26:05
26:19