Crypto is internet money and the internet is for robots.
We humans barely recognize their cogs and sockets — the crypto arbitrage and MEV bots, social media bot armies and algorithmic content feeds, generative code art and AI illustrations, and the various Zapier agents and automations running in the background of our experience.
In the past, I discussed how there will only be more robots, and that Web3 is the right economic and financial venue for their emerging machine economy. Of course, many of us use so much software that we also qualify as software cyborgs.
Lex Sokolin, the founder of Generative Ventures, is the former Global Fintech Co-Head at ConsenSys, a blockchain technology company.
Genative Ventures has been investing in groundbreaking companies since mid-2023, focused on the intersection of crypto, AI and fintech. Us The macro thesis was compelling enough to get started with, but there is reality on the ground and its patterns are starting to become clearer. Today we want to dive deeper into how this is already happening and describe the AI DePIN (decentralized physical infrastructure) trends are driving the sector forward.
On chain/off chain
The first observation is that some things happen on-chain, and some things happen off-chain. This is obvious, but worth mentioning. In the case of money, dollar bank deposits are off-chain, while DAI is on-chain. Real world tokenized assets are off-chain, wrapped tokens or liquid staking are on-chain. National passports and driver’s licenses are off-chain, while POAPs and NFTs are on-chain. The same concepts – money, financial instruments, identity – can be manufactured in different places.
The same goes for artificial intelligence. We could have a world where AI is off-chain, but occasionally switches to Web3 to take action. In such a case, we need services that function as oracles or on-ramps for machine intelligence.
The same logic that drove fintech to create embedded financial products (e.g. Plaid, Stripe) could give GPTs the API software tentacles they need to float around crypto markets. Who is the Moonpay for ChatGPT? And as we bring AI into our trusted environments, how do we keep it fair and verify its results?
Some teams in the market are working on inserting crypto technologies, such as ZK proofs, into the computational process of machine learning models. This would prove that a certain robot did indeed do what it was supposed to do, that you weren’t fooled by Bernie Madoff with a spreadsheet. Today, the approach is still in its infancy. However, in the future we can understand the value of verifying that you are communicating with the correct black box.
Others are thinking about how to move the entire LLM and neural network stack to a decentralized infrastructure. Because much of the generative AI movement is open source, such as crypto, it is conceptually possible to deploy and maintain the multitude of open source models on protocols that decentralize the computational burden, and create incentives to use the best provide machine intelligence. Services.
Despite several talented teams pursuing this strategy, it is still largely science fiction. Even centralized AI teams have yet to figure out the shape of demand and profitable unit economics. In our opinion, it is early to split the offering of such services into protocols and DAOs at this stage.
ThePIN defined
Lower on the stack is DePIN, a decentralized physical infrastructure. The simplest DePINs use coin protocol incentives, rather than more modern tokenomics approaches.
For example, participating nodes provide hardware storage, compute, GPU, or training data to a network and are rewarded for work done on their device, similar to Bitcoin proof-of-work mining. We think that most DePINs, like Helium, will no longer function like a coin, but will be more like a dApp running on top of a computational blockchain like Peaq, Solana or an EVM rollup.
Web3 used centralized cloud services to deploy decentralized networks, but we’re not sure if that will work for decentralized AI given the cost and demand. Centralized AI providers are simply more efficient, better organized and computer hungry than their protocol-first competitors. That’s why we think scalable DePINs would be a material enabler of on-chain crypto AI.
The other part of DePIN we like is that it integrates machines into Web3, and machines will need DeFi and its automated financial products, as well as access to intelligence as a service. In a distant future, decentralized car fleets can download the latest self-driving models from DePIN AI networks, maintained by various DAOs and incentivized by tokens. Small experiments in this direction already exist.
Furthermore, frameworks have emerged for connecting, standardizing, and managing populations of AI agents with different goals. Autonolas is one such project, which generates bots for participation in the trading market and is used in administrative procedures. If you want to understand a single agent, consider Botto, a generative AI model that produces art curated by a DAO and stimulated by a symbolic feedback loop. Or consider Numerai, a hedge fund running a token-powered competition for data scientists to build AI algorithms for a trading software brain. There will be many such beings – some simple, some complex and some inscrutable.
A final distinction we would like to highlight is the scope of AI services. In one scenario, it is only a limited function to improve applications. Take a digital wallet for example, which now offers the ability to talk about the tokens and investments it holds. Or a 10-K corporate filing on a website that you can query with a custom financial GPT. While this is useful, there has been no structural transformation in the sector.
Alternatively, there is a world where OpenAI becomes the new iOS and the GPT store is the new app store. The tech company then becomes the main conversational interface for accessing machine intelligence applications, which are embedded in its website. AI is not just a function, but the new platform that distributes solutions to a variety of common problems. In this scenario, crypto is accommodated as one of many AI functionalities.
In this case, one concern is that the AI agents are fundamentally centralized and housed in a single provider, creating enormous personal risks across data, privacy and finance. Custody always creates a principal/agent problem, where the agent has an incentive to steal from the principal, and therefore laws and regulations must enforce a fiduciary duty. In the big tech world, AI regulation that protects the individual in some form is inevitable.
AI agents and self-preservation
The Web3 counterbalance to this danger is the self-management of information and the self-management of our AI agents. Maybe we generate the GPTs on a centralized platform, but we can put those trained models into a crypto wallet for ownership. In a world where multiple successful open source models exist, and some run well on decentralized infrastructure, we expect crypto custody and control over AI agents to become a core value proposition of Web3.
Today, projects that guarantee the provenance of digital media in the AI era come close to this idea. Things have to be real and beyond manipulation.
Another example is NFT minters linked to image generation or LLM engines. This supports the fruits of machine labor with Web3 DeFi marketplaces. However, trade around such objects is still negligible today – whether this is due to the dire state of the NFT markets, the quality of machine labor, or the low utility of such digital assets.
In all cases, this is an absolutely fascinating design space for entrepreneurs. Since launching Genative Ventures, we continue to be amazed by the creative diversity and energy of technologists exploring the possible and charting a path forward.