The crypto industry is turning the page on a wave of high-profile scams and scandals, but that doesn’t mean the grifters are gone. In recent months, crypto hustlers have sought to hitch the industry to the shinier new artificial intelligence in hopes of parting the unsuspecting from their money. I’m here as your cynical guide to deciphering the specious business models that claim to unite these two unrelated fields.
It’s early days, but it’s a safe bet this new hustle will produce a dumb genre of crypto-tokens looking to capitalize on AI enthusiasm. As we’ve learned from past hype cycles, when all you have is a crypto thesis, everything looks like it needs a decentralized database slapped on it. We’ve seen that the lazy “AI plus crypto” efforts are poised to come in three different flavors.
The first and slightest offensive idea is using a cryptocurrency token to access an AI model or service à-la-ChatGPT. Part of this pitch will exploit the confusion over using “tokens” in both AI jargon and crypto-jargon. While the term “token” is used in AI models, rest assured that it has nothing to do with tokens in the sense of cryptocurrencies or blockchain-based ledgers.
In AI models, tokens are a word or a common syllable that can be thought of as a unit for processing – typically within a large language model (LLM). While some AI services will let users pay for subscriptions with cryptocurrencies, there is nothing special about crypto-tokens that makes them particularly suited to interact with LLM-tokens. Nonetheless, expect to encounter entrepreneurs looking to convince you otherwise.
Next, beware of attempts to rehash the “data is the new oil” meme that was everywhere in the 2017 crypto ICO boom. Grifters will be hawking tokens that reward people for contributing their data to an AI-training model (so-called “data marketplaces”). This made no sense in 2017 and will make no sense in 2023. If you have a presence on the Internet, your data is the horse that left the barn. There is no sensible way to monetize this asset, blockchain-based or otherwise. Anyone telling you otherwise is either in a Rip Van Winkle situation or simply doesn’t know how most services on the web work. Turnstiles are rarely placed in open fields.
Lastly, we should expect tokens to emerge with “AI” or “GPT” in the name (wait, we already have that!). Many cryptocurrency projects exploit the naivete of excitable people who make decisions based on light pattern association. Rest assured that some will distill that to its purest essence.
None of these “AI plus crypto” shoehorned tokens make sense to genuine AI developers. But the cryptocurrency industry has a habit of spinning up B-list business models in the hope venture capitalists will be gullible or greedy enough to repackage them. This includes rebranding existing crypto businesses in hopes of dumping tokens on retail suckers.
Look no further than Helium, a company founded as a wireless network for the “Internet of things” craze in the early 2010s. In 2019, with the help of dedicated crypto funds, Helium’s team pivoted to create a crypto-based model that would act as a “grassroots telco” enabled by purchasing physical hotspots that anyone could buy and receive network tokens (HNT) for operating. Eventually, the project’s reputation was marred by the shaky economics it offered for hotspot owners and evidence of spoofing.
Then there is Worldcoin, the Sam Bankman-Fried-funded, and Sam Altman-led gimmick to scan irises in return for crypto—part of an effort to bootstrap a basic universal income experiment and “identity network based on proof of personhood.” Or something. For those of us who can’t grow mustaches to twirl or lack fluffy white cats to stroke while laughing maniacally, paying poor people for their biometric data to become part of an operation with dubious security practices can only be categorized as pernicious bullshit.
It’s worth remembering that the principal value of cryptocurrencies is to offer a censorship-resistant store of value that unites people across borders. Fundamentally, cryptocurrency is accounting software. AI aspires to replicate human thought and amplify it. The two technologies exist on different ends of the spectrum of creativity for a reason.