10 Ways Crypto and AI Can Make Each Other Better (or Maybe Worse)
Poor crypto. Once the darling of the tech world, crypto — or blockchain or Web3 or whatever we’re calling it today — has been mired in lawsuits by the SEC; a frosty bear market; and now, perhaps most painfully, the indignity of playing second fiddle to AI.
Jeremiah Owyang, an influential entrepreneur who has been active in the Web3 space, lives in San Francisco and told me there are three to five AI events per day in his community, but there are “very very few Web3 events now.” (Owyang is now diving deep into AI.)
Jeff Wilser, a CoinDesk feature writer, is the author of 7 books including Alexander Hamilton’s Guide to Life, The Book of Joe: The Life, Wit, and (Sometimes Accidental) Wisdom of Joe Biden, and an Amazon Best Book of the Month in both Non-Fiction and Humor.
Sheraz Ahmed, Managing Partner at Storm (a blockchain consultancy), said that when he traveled to Dubai for a crypto event, “it was a bit shocking that in every blockchain keynote panel, AI was included. It was a blockchain conference, but only AI was being spoken.”
And there’s no question that the sloshy start-up capital is now enamored by AI. As Meltem Demirors, Chief Strategy Officer at CoinShares, joked on Twitter, “2019: we are a crypto fund. 2020: we are a DeFi fund. 2021: we are a blockchain gaming fund. 2022: we are a web3 fund. 2023: we are an AI fund.”
But maybe pitting Crypto vs. AI is the wrong framework? Web3 might be in “winter,” but it’s not dying. Projects continue to quietly build. And, if it’s true that AI will impact every industry, clearly that will include Web3. As CoinDesk’s Michael Casey recently wrote, “AI will need Web3, and vice versa. Why not build both?”
Projects like SingularityNET, Ocean Protocol, and Cortex have already been working on various types of decentralized AI solutions. Many more will inevitably follow. “I feel like there’s a lot training that you get from the crypto space that’s weirdly useful if you’re starting to think and engage about AI,” says Nathaniel Whittemore, who now hosts a daily AI podcast in addition to his daily Web3 podcast. By “training” he means getting accustomed to things like regulatory issues, educating others on a weird new tech, and how “you get used to being hated by some portion of the population.”
As for how AI and crypto will interact and intersect? “There’s a whole slew of possibilities,” says Whittemore. Some are obvious, some are outlandish, some are hopeful, and some are nightmarish. To impose a bit of order to the chaos, we’ll consider ten possible areas of overlap…which range from boosts in productivity to Armageddon.
1. AI will make Web3 projects more productive and efficient.
This is less about the unique properties of crypto and more about how AI will change everything, Web3 included. “I think AI will impact [Web3] in all the normal ways it does for other industries,” says Haseeb Qureshi, Managing Partner of Dragonfly. “Productivity will increase dramatically. Smaller startups will be able to do more.” He expects that Web3 projects will need to spend less on things like marketing, copywriting and other elements of operations.
Whittemore agrees. “AI will almost certainly radically increase the number of people who are capable of building blockchain and Web3 applications,” says Whittemore. And there could be a knock-on effect. He thinks that because it will become easier for non-tech people to learn things like how to code, that could potentially help “reduce the power gap” that’s ingrained in the community.
2. Blockchain could bring decentralized values, governance, and guardrails to AI.
This is the chocolate-meets-peanut-butter solution that many in the space have craved. “Blockchain platforms can support the creation of decentralized AI models,” says Ahmed, who adds that this solution would — in theory — keep data private, reduce biases in the models, and curb the growth of centralized monopolies.
“In theory,” of course, is the rub. There are skeptics. “[AI] is not a technology that lends itself to distributed systems,” says Qureshi, because they’re being built by small groups of experts and require a staggering amount of computing power. “They’re not the kind of thing where you can leverage a lot of decentralized resources, like you can with bitcoin mining.” (Many in the Web3 space, of course, disagree and are trying.)
Whittemore admits that he’s “enormously skeptical” every time he sees a new crypto/AI project, but adds that “there are a lot of really smart people thinking about exactly this,” and he imagines we’ll see more “interesting overlaps” in the second half of this year. He’s also encouraged that “open-source approaches [to AI] are outperforming everyone’s expectations,” which gives a jolt of optimism for a decentralized approach.
3. AI will upend crypto trading.
Crypto trading is 24/7, without borders, and it happens across a dizzying range of assets and markets. Even in the pre-AI status quo, humans struggle to keep pace with the always-on casino. That will get harder. Quant and algorithm-fueled trading “bots” have existed for years, of course, but they’ll get exponentially better. This could be lucrative for those who use them, death for those who don’t. The millions of “retail traders” could get squeezed. “AI is going to be the enemy of the stubborn-minded,” says Ahmed. “If a trader does not want to use [AI] and wants to do everything manually, they’re going to get left behind.”
4. Blockchain could save us from AI’s ocean of misinformation.
AI-injected deep fakes have already gone viral and tricked millions, and if the trend continues — which it almost certainly will — we could soon be living in a dystopian world that some have called a “post-truth society.” Reality will be up for grabs.
Crypto advocates have long argued that blockchain could be used to authenticate that digital assets — like photos and videos — are what they claim to be. “Zero-knowledge proofs do not get enough attention, but they will be extremely important for AI going forward,” says Joseph Raczynski, a technologist and futurist. “They will serve as the technology that helps prove something is something, without divulging anything sensitive.”
5. AI could make crypto compliance and auditing easier.
“Especially after the events of the last year, regulators are becoming more hands-on and demanding more compliance,” says Elbruz Yılmaz, Senior Vice President of Web3 and Crypto at Paysafe, a payments provider. Auditing is a manual and time-consuming process, says Yılmaz, and many Web3 companies aren’t set up to do it. They’re playing catch-up. “AI will help these companies get to a competitive level in compliance faster,” says Yılmaz.
6. AI could make life easier (or harder?) for crypto hackers and scammers.
The large language models of AI make it “easier to attack smart contracts,” says Qureshi. In the same breath, he adds that perhaps AI could also make smart contracts “easier to defend.” He doesn’t yet know which way that will tip — no one does. The only thing we can say with confidence is that AI will be weaponized by the bad guys and used as a shield by defenders; who wins is an open question.
Then there are the scammers. “Crypto scams are far more sophisticated with an AI involved,” says Pete Howson, a technology researcher and crypto skeptic, who’s also the author of the upcoming book “Let Them Eat Crypto.” “For example, ‘deep fake’ videos of celebrity endorsements are encouraging suckers to invest in the next big shitcoins.” This is not just speculation. It’s already happening. Howson continues: “In November 2022, a verified Twitter account posing as Sam Bankman-Fried posted a deepfake video offering FTX users ‘compensation for their loss’ in a phishing scam designed to drain their crypto wallets. Fraudsters are using AI platforms to build fake trading websites.”
7. AI could spur the growth of the metaverse.
With all due respect to Apple’s Vision Pro, the infrastructure for the metaverse is still a glitchy work-in-process. The space needs a better user interface. Better graphics. Better content. Better community. AI could help with all of that. AI will slash the time it takes to build immersive worlds, and it could even populate these worlds with human-seeming characters. Nvidia recently melted faces in the game developer community, for example, by showing how AI could power real-time dialogue with NPCs (non-player characters), which unlocks all kinds of potential for world-building. “AI can help these games become much more engaging, faster,” says Yilmaz.
8. AI could exacerbate the environmental concerns of crypto.
Okay, perhaps this one isn’t about making anything “better,” but we should acknowledge the concerns and risks of environmental impact. “Just like cryptocurrencies, generative AI platforms also use criminally huge amounts of energy,” says Howson. “The training of ChatGPT consumed 1,287 MW of electricity per hour, producing 550 tonnes of carbon dioxide. That’s equal to a person taking 550 roundtrips between New York and San Francisco.”
9. Cryptocurrency could be used by AI agents.
This is my favorite.
Qureshi imagines that in the future, as bots like ChatGPT get more advanced, they will essentially function as your personal assistant, or agents. They could book you flights, place your Instacart order, or summon your Uber at exactly the right moment. (Three years ago I interviewed a Web3 entrepreneur, Dele Atanda, who described exactly this scenario.)
Sure, some of this might fit in the current financial framework, but what about when my AI agent begins transacting with your AI agent? Or, what if the way AI agents spend money simply becomes too complicated for the laws or banks to catch up?
“They will need money,” says Qureshi, “And the idea that JP Morgan Chase is going to open a bank account for AI is nonsense.” Qureshi suspects that the “laws are not going to catch up in time,” but that crypto is already a functional solution that can be implemented immediately.
Whittemore says something similar. He gives a hypothetical: Imagine that you want to start an online store for band logo keychains, and the AutoGPT (or your AI “agent”) devises a set of tasks, and one of them might be to figure out how it’s going to 3D print the keychains. “Can it just send bitcoin to the 3D printing [AI] agent that runs autonomously in the background?” If the 3D printing agent is located across the globe — which is often the case — then bitcoin would be a more elegant solution than dollars.
So, in the end, after years of searching for “adoption,” the first mass-market use of cryptocurrency-for-transactions might come not from humans, but machines.
10. AI could leverage crypto to achieve the Nightmare Scenario.
We’ll zag with one more less optimistic possibility. (And “less optimistic” is putting it lightly.)
If you’ve been at all AI-curious, by now you’ve probably heard the cheerful stat that in a survey of AI researchers, more than half thought there was at least a 10% chance that AI could wipe out the human species. (Sidenote: This stat is slightly exaggerated and has been somewhat misreported, as the survey response rate was only 4% and not truly a representative sample of scientists. But still. Whether it’s a 1% chance or 10% chance, the “nightmare scenario” is still very much a scenario.)
And crypto could play a role in that nightmare.
“With greater dependence and importance on digital assets, an AI agent could alter a financial institution with ease, impact a governing body, or take down a government without anyone even knowing, until it is a bit too late,” says the futurist Raczynski, who also imagines that the AI could wreak havoc in a metaverse. Raczynski fears that the advancement in AI will allow it to soon surpass the abilities of humans, “and blockchain is an enabler.”
Final thoughts
So let’s conclude on something a bit less Armageddon-ey. Even though AI is now getting all the hype and the shine from VCs, the experts I spoke with were not concerned about the funding impact to Web3. “This is a silly thing to worry about,” says Qureshi, who acknowledges that an insane amount of money is pouring into AI “with sky-high valuations,” but doesn’t expect this to materially hurt the Web3 ecosystem. “There’s a lot of capital in the U.S.,” says Qureshi, who adds that no one can say with straight-face, “We don’t have enough money,” and that “I haven’t heard that in the last three years.”
And on a more hopeful note, ultimately, Ahmed thinks that AI and crypto needn’t be in competition, but are rather just “tools in a toolbox.” They’re complimentary. “They’re not fighting amongst each other,” says Ahmed. “Ultimately they’re going to be connected, just like electricity is used to power the internet.”