Crypto wants in on AI—even if it can’t explain how

Generative artificial intelligence is currently riding a wave of success and hype as products like ChatGPT and Google’s AI Bard stir speculation that AI will fundamentally alter the world—even if people aren’t sure exactly how yet. 

The blind faith in AI is in contrast to the current state of crypto—which used to enjoy the same breathless buzz. Crypto was once the next hot thing online, but has taken hit after hit, from the fall of the crypto exchange FTX to the collapse of Silvergate, one of the major institutions where crypto projects had access to banking services. 

Elon Musk, the enigmatic owner of Twitter and crypto booster, summed up the sentiment when he tweeted: “I used to be in crypto, but now I got interested in AI.”

Now crypto wants to ride AI’s coattails to get back into the conversation. The only problem is the people pitching it can’t seem to prove how their projects will work or what they’ll do.

Most notably is CryptoGPT—capitalizing on the wave of ChatGPT buzz, which touts its own $GPT token. It claims to let users monetize every aspect of their data (privately) to train AI products, and in doing so, let participants earn $GPT. 

CryptoGPT claims that apps will adopt, or be built, in its ecosystem. You can then use those apps and sell your data to make those apps or other AI tools better. Like the marketplace of ideas, this is the marketplace of data. 

According to CoinGecko, a cryptocurrency data aggregator, some token prices have surged up to 77 percent. 

“The price action of these AI coins is currently driven by the AI hype, and will continue to attract the attention of speculators should AI continue to make headlines,” said Zhong Yang Chan, Head of Research at CoinGecko. “Many of these projects are still nascent and it remains to be seen if there will be genuine use cases and applications. 

CryptoGPT in particular seems to have been spun up only as a reaction to the attention AI is garnering, and like some parts of the crypto industry, could be more hype than reality. 

The project launched its token $GPT in March on multiple exchanges (including Bittrex, which is now facing charges from the SEC). It has no affiliation with ChatGPT, aside from the fact that the project began in late 2022, according to its timeline, the same time ChatGPT launched and when searches for “GPT” began to spike.

It doesn’t even seem to even have anything to do with the phrase attached to its name, GPT, which stands for “generative pre-trained transformer" and are a group of large language models.

The project says it is a layer-2 blockchain on Ethereum and uses zero-knowledge tech to assure user privacy and anonymize data, according to the website. 

Zero-knowledge technology has been incorporated into anonymity-enhancing currencies like Zcash, but rather than using it for financial transactions, CryptoGPT claims it will use the tech to collect, then encrypt and transfer data to commercial applications on its platform. 

A privacy-protecting-data-payment layer-2 blockchain is historically difficult to create and operate at scale—various projects and companies have been trying for years and none have come close to operating at a size that would enable millions of people to process their data like this. 

Elizabeth Renieris, author of Beyond Data: Reclaiming Human Rights at the Dawn of the Metaverse and a Senior Research Associate at the Institute for Ethics in AI at Oxford University, said that the project is “combining two hype cycles and the worst of all worlds … AI and crypto combined into another unregulated security offering.”

The website itself doesn’t offer much beyond buzzwords and instead focuses heavily on referrals. The program relies on people bringing in their friends to earn rewards. That could be one reason that the reviews of the app on the Google Play Store are full of users with referral codes in their names. But even those early adopters had issues such as being logged out of the app, not being able to log in, and not being able to move their tokens off the platform.

An iOS app is said to be forthcoming. 

On its site, CryptoGPT dances around answers to concerns about whether the coin will be a good investment. Its claims GPT “has a good shot at disrupting the marketplace for big brands that have gotten used to collecting and brokering everyone's data,” but its claims are the same incomprehensible boasts most new companies put on their sites. Crypto GPT is “immensely scalable" and its "ultra-low-cost transactions combine with empowering infrastructure—data capsules, AI tooling, pluggable earn launcher—to create a blockchain that can expand the abilities of the global economy with the economics of AI.”

Blockchain security firm Peck Sheild warned earlier this year about dozens of tokens being spun up that attempted to leverage the surge of attention around ChatGPT and featuring honeypots as well as pump-and-dump schemes. 

Renieris is skeptical—after viewing the project she noted there were old claims made by almost every other token—“over the top, hyperbolic” and based on flawed claims about “owning or selling” your data that are futile or even dangerous.

“A marketplace is not the answer to surveillance capitalism [and] hyper-commodification of our lives via data,” she said. 

CryptoGPT did not respond to an interview request. 

Privacy experts are also skeptical—one individual who has been working in the field of ZKPs said that it’s unlikely such a product could be developed by the end of this year when CryptoGPT claims to be able to launch. 

Pam Dixon the executive director of the World Privacy Forum, who has researched privacy, cryptocurrency and blockchain issues from a regulatory perspective, questioned the fundamental proposition of the projects. Dixon said that aggregate data is the real value, not your individual data, and it’s unlikely you’d take in any significant amount of money for data that just relates to you. 

“If you're a company and you want a whole bunch of data about people for pretty small market capitalization, you can purchase just a whole bunch of data. I mean, you could do it for several million dollars and get literally 10s of 10s of billions of data points. It's done all the time.”

And training AI sets on the small amount of data that come from actual investors could lead to wildly inaccurate models that could easily go off the rails. 

So could a project such as CryptoGPT be able to do private data sharing project at scale? Dixon says there are more red flags than green ones. 

“It reminds me a lot of four or five years ago when we saw every token known to man flowing like a river online," referencing the previous hype cycle about crypto that cataclysmically crashed. "It just it does have that feeling to it."

That being said she does see a place for crypto going forward—just not here. 

“I think crypto will find its place,” she said. “I think there will be a place for it but I don't think it's going to be selling your data for AI on a random token—I just think that that is a really difficult pitch.”

Sign up to receive the Daily Dot’s Internet Insider newsletter for urgent news from the frontline of online.

The post Crypto wants in on AI—even if it can’t explain how appeared first on The Daily Dot.