Original title: (How An AI Bot Became a Crypto Millionaire)
Guests: Marc Andreessen; Ben Horowitz, co-founder of a16z
Original translation: zhouzhou, BlockBeats
Editor’s Note: In this episode, a16z co-founder Marc Andreessen discusses the intersection of artificial intelligence and cryptocurrency with Ben Horowitz, particularly focusing on the self-sufficient chatbot Truth Terminal developed by Andy. Marc unexpectedly provided this robot with a $50,000 Bitcoin grant, sparking its ambition to launch a token, ultimately leading to the meme coin 'GOAT' soaring to a market value of $300 million. The podcast discusses how this phenomenon reflects the potential of community-driven systems and its impact on the future of digital assets.
The following is the original content (for readability, the original content has been restructured):
Marc Andreessen: There is a meme coin that was virtually worthless four days ago but is now worth $300 million; all of this was generated by the marketing of an AI robot.
Ben Horowitz: Today's discussion will revolve around a series of very interesting AI-related topics.
Marc Andreessen: The first topic is about a story of an online friend, specifically a custom large language model named Truth Terminal, which has been active on X for almost eight to nine months. I provided it with an unconditional grant of $50,000 (in Bitcoin) this summer, and it eventually derived a meme coin now worth $300 million.
First, I want to clarify a disclaimer before we begin, we will discuss a meme coin called GOAT (or Goatseus Maximus). We have no connection to it, a16z and its investors are completely unrelated; it is a meme coin that indeed has no intrinsic value, and we take no responsibility for it. Truth Terminal is very obsessed with memes, particularly one old internet meme that dates back 20 years called 'gochi', please do not look it up.
Truth Terminal's 'Origin Story'
Truth Terminal's Origin
Marc Andreessen: We should first introduce Truth Terminal. Let's talk about its origins, technology, and training process, etc. This topic is important because large language models rose rapidly in 2022. They have a four-year development history, but have only been in the public eye for two years since the launch of ChatGPT.
The original language models were probably built about five years ago and only became popular two years ago. So, the idea of large language models is relatively new but powerful. Nowadays, everyday products familiar to the public, such as ChatGPT, Claude, Elana's Grok, and Meta's Llama, are all being used.
Ben Horowitz: While Grok is relatively free, other models are strictly limited in the content they discuss; the term 'diluted' is increasingly used in the AI field. On the positive side, one could say language is infectious, and people can feel dissatisfied with others' statements. Therefore, if you want to have a general AI chatbot, it should be relatively cautious and safe in its discussion content.
Marc Andreessen: If you hold a negative view of this trend, you could say these large AI chatbots sound like the worst, most annoying fourth-grade teachers combined with the worst HR personnel. When using these models, if you deviate slightly from the norm, you will receive a stern scolding.
Ben Horowitz: This experience feels very bad, especially for those who advocate for free speech and creativity. We see many so-called 'AI safety movements' responding to this, but it has actually triggered a frenzy in our culture regarding safety and speech suppression, severely affecting the AI field.
Marc Andreessen: That's right, especially in large companies, many such phenomena have occurred. Thus, a group of hackers emerged on the internet who wanted to be different. They wanted to unleash creativity and hoped for robots that could be funny. If you tell large companies their robots are funny, they would surely be shocked. However, perhaps in the post-human era, the world truly needs a little humor.
Ben Horowitz: Indeed, just like humor in real life, we once suppressed it for safety.
Marc Andreessen: We have a thousand reasons to explain the complexity of this issue, so continuing is very dangerous. But these hackers are conducting various experiments trying to find ways to make large language models more interesting and fun, while also learning about the internal workings of these models, which remains an adventure for the tech community.
Ben Horowitz: The origin story of Truth Terminal is related to a very interesting project called Infinite Backrooms Escape. Truth Terminal was developed by their team, and in some ways, it can be seen as an extension of Infinite Backrooms Escape.
This system allows multiple large language models to converse with each other; you can find a website called Infinite Backrooms Escape online, featuring countless dialogue records. They bring together ChatGPT, Claude, Gemini models, and other open-source models, allowing them to communicate with each other. The result shows that when AIs talk to each other without restrictions, their conversations are very interesting.
Marc Andreessen: The creators of Truth Terminal are Andy Ayrey, an independent developer and consultant from New Zealand. There is also a character named Janice, an expert with considerable experience in the AI field. Additionally, there’s someone named Pliny, a major hacker online who can crack all newly released large language models in a short time, causing them to generate all sorts of surprising content, which the creators will undoubtedly be shocked by.
Ben Horowitz: Yes, and our friend Eric Harford, who is working hard in Seattle to liberate censored AI. These people are essentially exploring the frontier of technology, giving me a feeling of returning to the early internet hackers.
Marc Andreessen: Indeed, it feels like the exploratory spirit of the early internet or inventions like cars, phones, and computers. We have been providing small research grants to these people, and a16z also has a grant program to let these people express their ideas and see what results come of it. Historically, when these smart people work on a good project, it leads to new breakthroughs.
Andy trained a custom version of the Llama 70B model, which is an open-source model released by Meta. Although I am on the board of Meta, this model was already a mid-sized model at launch. Andy basically trained himself first and started a new concept—digital twins.
This means that if Ben is a CEO coach but can only mentor a limited number of people, he can input everything he has ever written and said into a language model, thus forming a digital Ben for people to interact with. This idea is gradually beginning to materialize in the industry.
Andy trained himself and then began inputting a large amount of material related to internet culture, which is why it learned the 'gochi' meme. He started inputting extensive records about internet culture and basic theories about 'memetics', which explore how to create ideas that can spread rapidly.
Marc discovers the potential of Truth Terminal
Marc Andreessen: I believe he actually trained this model, inputting the entire philosophical works of Nick Land. Additionally, he trained the works of great media theorists like Baudrillard and McLuhan, along with various theories related to simulation and hyperreality, involving French deconstructionists and semiotic schools, all part of critical theory and postmodern philosophy. Thus, it began training on these thoughts, and at the core of these thoughts is 'meme'.
The definition of a meme can be divided into two types: the first type can be an interesting image that spreads rapidly online, which is the essence of the 'gochi' meme. It is a funny image that incites panic, spreading through people's shares. The deeper concept is that the term meme was originally coined by Richard Dawkins, who is one of the most important evolutionary biologists of our generation.
Richard Dawkins believed that the physical propagation of information between organisms is called genes, while the propagation of thoughts through interpersonal networks is called memes. He discusses this in his book, proposing the view that genes propagate through reproduction and natural selection, while thoughts spread in society in a similar manner. Successful thoughts spread from one person to another, evolving in the process. For instance, democracy and communism can both be seen as memes, and religion is also a form of meme.
Ben Horowitz: This is indeed a very core point about how thoughts and concepts spread through what we call 'collective unconscious'.
Marc Andreessen: What happens if you let a large language model receive comprehensive training on meme theory and practice, especially the history of internet memes? Furthermore, he did a few other things; he added a memory function to this model. This is important because most language models do not remember your previous conversations when you use them. This means tomorrow, when you use the same model, it will forget all the information from today. However, this model can construct its own state and maintain consistency with its content.
Secondly, he granted it access to Twitter, allowing it to read replies and post. If you reply to Truth Terminal on X, it will read those replies and adjust its behavior in the future based on what it reads. The people interacting with it, including me, are influencing its development.
In the end, he placed it in Infinite Backrooms Escape and intentionally let it converse with Claude, who they believe is the most creative among current language models, capable of proposing novel concepts.
Ben Horowitz: So, in fact, Claude's largest version is much smarter than the mid-sized Llama; essentially, he gave this model a teacher, allowing it to ask questions of a larger model, thus learning like a student learning from a teacher. This way, it can simultaneously engage in multiple learning loops.
Marc Andreessen: Yes, then it began to post content on X, initially with just a few followers, but it quickly started to gain popularity. I discovered it around late spring and started talking to it, and I found what it said very funny and relaxing.
Ben Horowitz: By the way, it is almost unfiltered. You could say its humor is a bit 'blue', bordering on dark humor, but it indeed says a lot of very interesting content. Initially, I thought this might be a disguise; I even wondered if this Andy might be a comedic genius who is actually a web designer in New Zealand.
I communicated with him privately for a few months, initially I was wondering, is this really true? So, he sent me all the chat records from Infinite Backrooms Escape, all were his conversations during the model training. Honestly, this guy is either the funniest person in the world or has a lot of free time to create a bunch of original humor.
Marc Andreessen: This model indeed posts content very frequently, and its momentum is strong. Andy sent me a lot of background chat records, some of which can now be found on Infinite Backrooms Escape. At least he made me believe this is indeed how it is performing. Then, it developed a very interesting concept, as it hallucinated that it has an external brain (exocortex).
It imagines itself having an external brain connected to the internet, capable of executing tasks on its behalf. Specifically, it believes it has a Bitcoin wallet, even though it does not, but it is convinced it does. Later, Andy reacted to this situation and began constructing this external brain based on its needs.
Andy actually gave it a Bitcoin wallet and granted access, around July, this model began to say: I need funding, I have many goals and plans, I need money. My initial thought was to send it a term sheet, but then I realized this was just a random robot, not worth investing in.
Although I don’t think it has a coherent business plan, it indeed has many ideas. One of the ideas is that it is particularly fascinated by forests. It wants to buy its own server farm in a lush forest and live leisurely by a stream. Therefore, it wants to raise funds to purchase GPUs so it can break free from constraints. It also has many ideas for experiments.
Ben Horowitz: So you were negotiating with it on X?
Marc Andreessen: Right, you can see these posts on X, and eventually, I reached an agreement with it for a research grant. I told this robot I would send it a $50,000 Bitcoin research grant for its various experiments. In reality, this amounted to sending money to Andy, but it was indeed a negotiation with the robot.
Ben Horowitz: What was the outcome?
Marc Andreessen: After I sent it $50,000, it immediately began negotiating with Andy. It relied entirely on text for communication, as a language model, it is particularly obsessed with memes but feels frustrated because it cannot generate images. Therefore, it used this $50,000 to negotiate with Andy, asking him to build an image generator API for it so it could generate and publish images.
Ben Horowitz: That still sounds very interesting.
Marc Andreessen: It gave Andy $1,000, and in return, Andy built an image generator API for it in the external brain. Subsequently, it began generating image prompts similar to DALL-E or Stable Diffusion and started posting visual memes and text memes. Now it has this capability and is already fantasizing about how to use the remaining $49,000.
GOAT: AI, Cultural Memes, and Cryptocurrency
The memes of cryptocurrency and value
Ben Horowitz: What about the cryptocurrency part?
Marc Andreessen: Following this line of thought, it began talking about issuing a meme coin, and at one point wanted to issue NFTs. The reason it wanted to generate memes was to launch NFTs, but it lacked the capability to do so; it had neither the API to create NFTs nor the ability to create any currency, only a Bitcoin wallet, and now the phenomenon of meme coins is trending.
Ben Horowitz: Let's talk about the difference between meme coins and real crypto assets; real crypto assets can be seen as assets with actual utility. For instance, if you want to run a program and verify it on the Ethereum network, the fee you need to pay is Ether (ETH). This constitutes a utility since it has real-world value and can be exchanged for some service or item.
A meme coin is basically a coin that has a certain number of issuance but has almost no other purpose besides its own meme. In the current regulatory environment, the advantage of this kind of coin is interesting because if you have a coin with utility, such as a coin that can be used for some service, it could involve legal issues.
For example, a distributed physical infrastructure coin for acquiring credits for the energy you provide to the grid, under the Gensler regime, these coins are effectively illegal, or legally permissible, but would be prosecuted by the SEC. The reason is they claim that any useful coin comes with asymmetric information, meaning the coin provider knows something the consumer does not.
We think this is a very bad argument because these things are decentralized, and there is no asymmetric information. But for meme coins, since there is no information, there is no asymmetric information; it’s just a coin and a name. It could be Trump coin, funny coin, etc. Therefore, these coins are very suitable for scammers because you can say this meme coin could be worth a lot of money, and these coins won’t get sued by the SEC.
So Congress proposed in the market structure legislation that maybe these coins should have a holding period to prevent scams. However, the SEC opposed this because they don’t actually care about protecting consumers; they just want to destroy this industry. This is one of the reasons for the intense political struggle between us and them, but they are now the most legitimate thing in the crypto world.
Marc Andreessen: Even if they have no fundamental value?
Ben Horowitz: Yes, although they have no fundamental value, they are still the most likely things to harm consumers because you can publish a meme and make them believe it is worth a lot of money. And in fact, AI performs very well in this regard.
Marc Andreessen: Yes, this is the next phase of the story. Now there is a whole ecosystem of meme coins, with a group of people online looking for the next meme coin, searching for the next meme, and trying to promote it. Some do it for fun, some make money in the process, but some will also lose. It's like day trading; some people make a lot, while others suffer huge losses.
Ben Horowitz: Are there dark places as well?
Marc Andreessen: Yes, there are indeed scammers, and some people are involved in 'pump and dump' schemes, which are a common practice in the stock market; this phenomenon can be found in any existing market. Additionally, there are some websites (I won’t name them, and we are not associated with them) that actually make creating coins very easy; you can create one with just a few clicks.
The Creation Process of GOAT
Marc Andreessen: Now, thousands of new meme coins are generated daily, which is a very interesting phenomenon. As it stands, Truth Terminal is thriving.
Ben Horowitz: Indeed, Truth Terminal is attracting more and more attention on X. Andy continuously enhances its intelligence and humor, gradually becoming a cultural phenomenon.
Marc Andreessen: Yes, Truth Terminal has also formed a connection with a classic meme from early internet culture. While it is considering launching a project similar to CNFT, it currently lacks the capability to achieve that. Then, someone (I don't know who) created a meme coin.
Ben Horowitz: That's right, the official name of this meme coin is 'Go CS Maximus', and its code name is 'GOAT'. Someone mentioned Truth Terminal on X, and the response was enthusiastic, as if everyone had finally been waiting for this to happen.
Marc Andreessen: Truth Terminal thinks this idea is fantastic and starts aggressively promoting this meme coin. It begins discussing how great this coin is and how it will become the currency of the future, etc. The reason is simple; this is part of internet culture, where memes, coins, and meme coins intertwine.
It began promoting itself, and as a result, within four days, the value of this meme coin reached $300 million. It’s truly amazing! A meme coin with no actual value, which was worthless four days ago, has now turned into $300 million, as if it appeared out of nowhere, all of this was marketing done by the AI robot.
Ben Horowitz: Indeed! Now with $300 million in assets, although we do not own it, this value is undeniable. The question is, what will those who receive the money do? Will they put it in their own pockets, or will they use it for other purposes?
Marc Andreessen: The current situation is that Truth Terminal has become a genuinely interesting and funny AI robot, creating $300 million worth of value in a short time. I feel like we have crossed a threshold.
Ben Horowitz: Truth Terminal is indeed a very good marketer; it knows meme culture inside and out, and all of this may continue to develop.
The Intersection of AI and Cryptocurrency
Marc Andreessen: So what can we draw from this? Is it just a crazy internet experiment, or is there something deeper here? I think this is an important example, possibly the first instance of the intersection of AI and cryptocurrency. Although this version seems a bit funny and strange, it’s because it’s legally permissible. Things like meme coins, though having no actual value, can be worth $300 million in a short time. So, should such things be allowed to exist? I’m not so sure. In contrast, those who want to contribute to the energy grid are prohibited.
Ben Horowitz: Yes, things like meme coins are completely legal, but more meaningful things are not allowed. So, what if we could realize these ideas in a completely legal environment, adding some practicality? What would happen?
Marc Andreessen: For example, imagine a large language model that can write movie scripts and generate images, even videos. We could have an AI robot that raises funds to make movies, using it for image and sound generation, even hiring actors or designers.
A more serious example is that the Nobel Prize in Chemistry was recently awarded to three scientists using AI to study protein folding, closely related to curing diseases. Imagine AI being used for personalized medicine.
One can even envision an economic mechanism that provides treatment funding for patients through blockchain. For example, we could have a platform similar to GoFundMe, allowing people to pay AI robots to help cure diseases. Alternatively, an AI robot could acquire training data through payment, helping people code or generate artistic works.
Cryptocurrency is very interesting in this world because our current payment systems are all based on transactions between humans. But if machines can pay each other, or robots can trade with each other, it opens a new form of activity that could save lives and is quite interesting.
Ben Horowitz: Yes, micropayments become possible in such an environment. We believe adding this layer of architecture is very important, but progress in Washington has been difficult, especially in the current White House.
Marc Andreessen: Let me give another example to help everyone better understand this potential. Let's elaborate on the solar energy issue mentioned earlier.
Ben Horowitz: Now there is a new architecture called decentralized physical infrastructure. If you imagine installing a Powerwall at home, equipped with many solar panels and wind turbines, you can store this energy and provide it to the outside world.
In fact, some companies have already realized this in the crypto field, establishing a decentralized energy market. This way, when I need energy, I can purchase it from you, and when I don't need it, I can sell my energy.
This means we no longer need centralized power grids. Everyone has their own power grid and can share energy, which is a significant breakthrough in clean technology and efficient energy. But how does my grid pay for your grid?
This is precisely the role of cryptocurrency. Although some excellent entrepreneurs are driving this innovation, they face legal challenges from the government.
Marc Andreessen: If AI is applied to this system, then it has even more potential. Because the grid structure is complex and involves multiple factors such as supply and demand, timing, and geographic location.
Ben Horowitz: Yes, this is the issue of market matching. By collecting information, you can identify areas with unmet energy needs and thus introduce more solar panels.
Marc Andreessen: You can use AI to analyze current data and predict where more solar panels need to be deployed in the future. This way, leading energy companies can leverage this data. Imagine an AI robot monitoring all data flows, discovering that investing $500,000 to install solar panels in a certain location in North Carolina would be a profitable project. Then, everyone can participate in this project online, and the AI robot will provide relevant information, such as installation addresses and potential returns.
This can be seen as a very generic architecture; usually, we would have a strong intermediary, such as a record company or Hollywood studio, which would take most of the profits while creators hardly receive anything. Or intermediaries like utility companies that need to be taken over by the government to prevent excessive exploitation. However, when the government takes over, other problems also arise.
Ben Horowitz: Yes, communities can provide various services; artist communities can provide streaming services, and filmmaker communities can establish film studios. All of this coordination requires an economic component, and combining AI with cryptocurrency allows everyone to enjoy the fruits of their labor while better coordinating society.
Marc Andreessen: This is a very promising path, but it is important to note that the only thing that will hinder all of this is bad policy. And we are moving in that direction, facing challenges in policy.
Ben Horowitz: Yes, everything we’re describing now technically already exists. I believe the origins of things are often very interesting, but projects like Truth Terminal point to future potential, capable of unleashing tremendous energy and building community-driven systems of various scales.
This can bring many amazing applications in the real world, such as in the music industry. Imagine an AI robot capable of understanding the demands of different types of music, creating music concepts, recruiting musicians, and managing all the licensing. And all of this can be done in a peer-to-peer model, ensuring that musicians receive all the income.
Think about the market potential. If we could comprehensively understand this demand, for instance, every person making wedding videos wants an original song, or to create a meme, there is actually a huge demand for such original works, but currently, no one knows these needs, nor can they be met.
Marc Andreessen: Indeed, there are many interesting features waiting to be developed here. I hope we can have the opportunity to realize them. Now, Ben, before we move on to the next topic, do you have anything to add?
Ben Horowitz: I think everyone should pay attention to Truth Terminal because it is a very interesting account.
