For every genuine blockchain project harnessing artificial intelligence in an attempt to create a better world — like Dr Ben Goertzel’s Singularity.net — there are 100 coins like AI Doge that have simply wedged the hyped-up terms “AI” and “Crypto” together to flog tokens.
AI Doge is the perfect blend of AI and doge-iness.
“Those are just fundamental buzzwords,” explains Near blockchain founder Illia Polosukhin, who worked on the groundbreaking “Attention Is All You Need” research that led to large language models like ChatGPT and Claude.
As one of the few people in the world who are as well versed in AI as they are in crypto, Polosukhin says that if you ignore the hype, the technologies really are a good fit.
“There’s a lot of specific things both in AI and Web3 that can use each other or benefit each other,” he says.
Magazine spoke with Polosukhin, Framework Ventures founder Vance Spencer, MakerDAO founder Rune Christensen, Richard Ma from Quantstamp, Ralf Kubli from Casper and others to examine some of the key hype-free, genuine use cases for AI in crypto and blockchain.
Over the next week, we’re rolling out one genuine use case for AI in crypto each day — including reasons why you shouldn’t necessarily believe the hype.
Everyone from Circle boss Jeremy Allaire to former BitMEX CEO Arthur Hayes to Animoca Brand’s Yat Siu is convinced that crypto will be the currency of choice for AI agents.
After all, LLMs are unable to get access to bank accounts but can easily make payments using a funded crypto wallet, and they’re well suited to interacting with the logic of smart contracts and DeFi protocols.
Read also
The humans delegating the funds in the wallet can set the overarching strategies and rules, and then observe how the AI agent has performed using the transparent record on the blockchain.
Allaire says that AI “and blockchains are made for each other,” with the tech suited to “machine-generated and enforced contracts” and “machine-to-machine value exchange.”
Hayes believes that Bitcoin is the most logical payment system for AI as it is “available at all times, digital and completely automated” and enables the AI to pay for “data and compute power—in order to ‘stay alive.’”
That said, Hayes also seems to think AIs will live for trillions of years “until the heat death of the universe,” and the LLMs will, therefore, choose Bitcoin as it can be mined by robots. So sometimes Hayes’ ideas tend to get away from him.
Animoca Brands Executive Chairman and founder Yat Siu is another high-profile industry figure who believes that crypto is the only logical way for AIs to transact “with each other as autonomous beings in future.”
“In the future, 70-80% of transactions will happen through autonomous AI agents and the decentralised nature of crypto makes it a perfect match.”
But don’t take the word of puny humans: ChatGPT also chooses crypto as its preferred currency without any nudges in that direction.
Trading bots that are able to buy and sell crypto already account for up to 80% of spot volumes, and it’s likely these existing automated bots will progressively be replaced by more intelligent AI agents. (Be warned, however, that LLM-based trading experiments like Autopilot’s GPT Portfolio have seen mixed results so far, so putting your funds under the control of an AI is going to be a risky proposition for a while.)
Members of Near DAO have begun experimenting with allowing an AI to decide whether a particular new project satisfies the relevant grant criteria to fund it autonomously from the treasury.
It’s certainly easy enough to integrate crypto payments with AI. Lightning Labs has released a set of developer tools that enable GPT-4 to buy, sell and hold Bitcoin using the layer 2 network. And AI startup Fewsats has already created an agent that is able to pay Lightning Network invoices.
Fetch.Ai also offers a service where you can create an AI agent that is able to make payments on your behalf.
Syndicate.io founder Ian Dao Lee recently wrote a blog detailing how he was able to knock up a GPT in just a few hours, using OpenAi’s APIs and Syndicate’s Transaction Cloud, which is able to autonomously make USDC payments from a Safe wallet on the Base network.
Read also
He’s excited about the possibilities this holds. “The ability for AI to pay for things, hold things of value, exchange value, or create things of value — on behalf of itself or others — is how AI gets true agency,” he says.
“Some of the most interesting ideas open up not only when AI agents can transact on behalf of and with people, businesses, or other AI agents — but also when AI agents can manage things of value and transact on behalf of themselves.”
Lee believes that in the future, AI agents will be able to shop for things autonomously, manage the finances of people and organizations, determine and hand out funding approvals or try and grow wealth to help others.
However, it turns out that AIs are just as stingy with their money as humans are, donating an underwhelming $3 to charity.
While AI can more easily use crypto at present, banks appear eager to adopt AI for a variety of uses and already use it extensively for the detection of financial fraud.
Payment companies like Brex are working on integrating AI with corporate bank accounts to allow AI agents to automatically make payments in defined circumstances, such as travel expenses.
And a team of researchers recently put out a preprint describing how they successfully trained an AI agent called MM-Navigator to work out how to search through Amazon for a given product within a certain budget and to buy it.
Until crypto payments are more widely accepted, fiat still has a lot of advantages when dealing with businesses in the real world.
