Almost exactly a year ago, I wrote an article for CoinDesk called “10 Ways Crypto and AI Will Make Each Other Better (Or Possibly Worse).” At the time, the idea of ”crypto + AI” was still a novel concept. On the fringe, too. To many, AI development felt, at best, adjacent to the Web3 world. To others, “crypto + AI” felt like a cynical combination of two over-hyped trends, like VC Mad Libs.
Fast forward a year later.
Decentralized AI is perhaps the hottest topic in the Web3 space. Dozens of projects (probably hundreds, it's hard to keep track) are currently competing to marry blockchain tools with the taming of AI. The space is booming: by the time you finish reading this article, another DePIN startup will have launched. And the architects of the movement will be gathering in Austin for CoinDesk's inaugural “AI Summit” in the run up to Consensus 2024. (Not to be biased: I helped organize the event and will be hosting it on May 31st.)
So why is this happening and why does it matter?
One could argue that the risks of decentralized AI are higher than those of decentralized finance. In some ways the principles are the same. Satoshi Nakamoto was wary of centralized financial institutions doing bad things and putting people's money at risk. As AI becomes a bigger part of our lives, and it feels inevitable, it will have the power to shape how we see the world, how we interact with it, and how we present ourselves to it. It could shape the way we think. Should this be controlled by billionaires?
Let's look at two concrete examples, one that's easy to understand and one that's scary: The next frontier for Gen AI will be the rise of AI agents that can book flights, pay phone bills, invest in stocks, and more. (“Hey Siri, go ahead and deposit $1,000 into the S&P.”)
Do we want Big Tech to control not only our data but also what we do with it?
Even scarier is the possibility that millions of people will one day use AI chatbots as coaches and therapists, perhaps via audio or video. An initial version already exists. (For example, on my podcast “AI-Curious,” I interviewed a veteran coach who is developing AI-based personal coaching in collaboration with a German research center. It's coming. It's inevitable.) Millions People begin to share their deepest thoughts, longings, fears, sexual desires, confessions, embarrassment, etc. Do we want Zuckerberg et al to chop this up to pieces? Decentralized and privacy-preserving blockchain technology has the potential to reap the benefits of AI without being poisoned by Big Tech.
These are the risks of decentralized AI. These are the topics covered at the AI Summit. (Or some of the topics. Others include: How blockchain preserves truth in the age of deepfakes; Fighting censorship; How to make “AI computing” more inclusive; How to fight bias; (Including ways to connect AI.)
While there has been an explosion of AI + crypto projects since 2022, some have been around and quietly building for years. One of them is Singularity NET, founded in 2017 by Ben Goertzel, a prominent AI researcher who had an early vision of building a decentralized AI. “The initial vision for what would become SingularityNET came to me in 1995 or 1996 when the internet came out,” Goertzel said recently at an AI conference hosted by Singularity in Panama City. “It seemed to me that the internet shouldn't be centralized. The internet is a decentralized network. So why do we need a centralized search service?” The seeds of this idea include AI (which he has been studying since the '80s), and Goertzel has proof. All these concepts coalesced in a book published in 2001 called “Creating Internet Intelligence.”
Goertzel, along with other pioneers trying to figure out the plumbing, pipes, and protocols of decentralized AI, will share his thoughts on how to make this vision a reality at the AI Summit, where he will talk about DePIN, a decentralized physical infrastructure network where a community of token holders will be encouraged to participate in building online services.
Gensyn, for example, is building a decentralized network that can harness the untapped computing power of individual users (like you and me) for training AI data, like Filecoin for cloud storage. “We’re running out of places to build huge data centers,” Gensyn co-founder Ben Fielding told me last year, before this was even a trending topic. If Gensyn pulls it off, “instead of having one huge data center, you’ll have every data center on the planet,” Fielding says. (Fielding will also be on stage at the AI Summit with the heads of IoTeX and Grass, the DePIN projects.)
While many of the Web3+AI projects are infrastructure efforts, there are also currently attempts to build decentralized GenAI models that directly compete with OpenAI around the world. And that brings us to another example of the convergence of cryptography and AI. Eric Voorhees is a Bitcoin OG, vocal supporter of Morpheus (a decentralized open source AI), and recently announced the launch of VeniceAI.
In a thread on Twitter/X, Voorhees said Venice is similar to centralized chatbots like ChatGPT, Claude, or even Grok, but “without any Orwellian elements” and that “Venice doesn't monitor.” , claimed that “Venice does not censor.'' Your Conversations,” and “Venice does not inject bigotry, safetyism, or political propaganda.” Just 10 days later, he showed it in action and added, “Venice.ai has no ads and a basic “I get the same answer as MSFTCopilot.” he fired.
Voorhees and Venice are just a few of the players in this field. There are countless others. New launches and new partnerships emerge every day. For example, in the last few days, Eternal AI announced his partnership with Filecoin. The Twitter/X post explains: “Filecoin provides decentralized storage for AI models and Eternal AI provides decentralized storage with a network of AI agents” reasoning. “
Is there one reason why these projects have suddenly taken off? To be a little cynical, there is certainly heat in this space, and VCs love heat. (The resurgence of the crypto bull market doesn't hurt.) But another reason is that, in a sense, there is no other choice. There aren't that many centralized AI players. The costs are staggering, so only a handful can be played at the table.
“Concentration of power” [of AI] said Nathaniel Whitmore, aka NLW, host of the AI Daily Brief and founder of BeSuperAI, which hosts AI town halls on consensus. “The more focused you become, the more you feel the need for a decentralized approach.”
None of this will happen overnight. In some ways, that is still a long way off. We need the smartest people in the field to collaborate, discuss and exchange ideas and inspiration. We need an AI summit.
