The Blockchain Bubble will Pop, What Next?

Last week, I flew from London to Tel Aviv. The man sitting to my right was a road warrior, just this side of a late-night bender in London. He was rocking an ostentatious pair of headphones and a pair of pants ripped wide apart at both knees. Perhaps a D.J.? At some point, circumstances emerged for us to commiserate over the experience of flying on Easyjet (not the easiest). Soon after, we stumbled through the obligatory airplane smalltalk: Where are you going? What do you do?

Turns out I was flying next to the CEO of an AI+Blockchain startup.

This image ran in an article in the Express discussing conspiracy theories suggesting that cryptocurrencies were invented by an advanced artificial intelligence.

It’s always a bit surreal when I learn of entrepreneurs combining AI with blockchain technology. For the past few years, whenever I found my myself bored among Silicon Valley socialites, this was my go-to satirical startup. What do you do? Startup CEO. What does your startup do? Deep learning on the blockchain… in The Cloud. Whoa.It was the poorest-conceived, most transparently buzzy nonsense I could think of. The formula is simple. People are excited about A. People are excited about B. What do you do? A + B. Done.

I wish I could have extended the benefit of the doubt. Perhaps he was a comedian? I was due to be on the receiving end of this joke. However, over the past year, buzz about AI + Blockchain startups has become commonplace. This buzz comes both from predictably unreliable sources like Futurism, which pronounced AI Blockchain likely to be the most disruptive technology of the year. It also comes from more respected sources like TechCrunch, and GeekWire, which have increasingly posted ingenuous articles corroborating that “AI on the Blockchain” is an actual industry. IBM even announced a product combining AI with blockchain technology to verify the identity of precious gems.

In my short conversation with the AI+Blockchain CEO, I tried to get a read on what his company was doing. While I am well-versed in machine learning, I don’t purport to be an authority on blockchain technology. Perhaps I was overlooking something real?

Chatting between sips of stale coffee,  I collected enough data points to discover he too had no great expertise in cryptography or fintech more broadly. As the conversation went on, I gleaned the following. The company works with smart contracts. Their customers include government officials, many of whom are nominally excited about blockchain technology but lack technical literacy. And his company is making lots of money.

This part was delivered with a mixture of impishness and disbelief. I imagine conversations like this must have taken place during the actual gold rush. I don’t know what you’re doing, or how well they’re paying you, but there’s literally gold flowing in the river. How are you not getting in on this?

Then he explained that AI was the big differentiating factor. Naturally, this caught my attention. In the remainder of our conversation, I tried to pry from him just what role AI was playing in their product. It seemed like the role of AI in the smart-contract was to determine whether terms in arbitrary real-world contracts had been met. When I pushed for specifics of how this could actually work, of how a supervised learning system trained on historical data, could be trusted, out-of-sample, to recognize that arbitrary conditions, perhaps not represented well in the training data had been met, with sufficient confidence to execute irreversible contracts, I got nowhere.

The answers ultimately led me to believe that the CEO of a venture-funded AI-based smart contract company knew markedly less than a competent undergraduate CS student about machine learning. Optimistically, it’s possible that he was the salesman  in a 2-person (or larger) founding team. Still, the existence of venture capitalists and institutional customers susceptible to this pitch strengthened my pre-existing belief that hysteria over the blockchain is unsustainable and likely to burst in the near future.

There’s a famous story about Joseph Kennedy and the stock market crash of 1929: it’s said that he recognized that it was time to exit the stock market because shoeshine boys and taxi drivers started giving him stock tips.  While I find the classicism of the telling tasteless, it may contain a nugget of truth. Bubbles are characterized by the marriage of excitement and ignorance, with markets flooded with eager but naive actors motivated by the perception of momentum.

This pattern of enthusiastic ignorance repeated during the dot-com bubble, when entrepreneurs popped out of the woodwork. Many had no engineering ability. Some had learned only how to register a dot-com domain and markup a website with HTML. Before 1999, the lack of qualifications didn’t seem to faze investors eager for exposure to the emerging internet technology sector.

The blockchain boom is characterized by a massive interest in cryptography-based technology — and yet too often, neither the investors specializing in the market, nor the founders of crypto-based ventures seem to possess a coherent vision, or even expertise. While I’ve suspected this for a long time, my brief contact with the AI+Blockchain world affirms that a significant portion of the blockchain boom has no legs.

As with the stock market crash of 1929 and the dot-com crash of 1999, I suspect the blockchain bubble will explode. However although the dot-com crash walked back valuations and chastened investors, a mature climate of internet businesses emerged in its wake. Today, technology firms account for 7 of the 10 largest companies in the world. The market didn’t overestimate the Internet, only the current crop of entrepreneurs. The question remains open of what impacts of blockchain-based technology might emerge long after the clowns have departed.

Related Posts

From AI to ML to AI: On Swirling Nomenclature & Slurried Thought

Press Failure: The Guardian’s “Meet Erica”

The AI Misinformation Epidemic

Author: Zachary C. Lipton

Zachary Chase Lipton is an assistant professor at Carnegie Mellon University. He is interested in both core machine learning methodology and applications to healthcare and dialogue systems. He is also a visiting scientist at Amazon AI, and has worked with Amazon Core Machine Learning, Microsoft Research Redmond, & Microsoft Research Bangalore.

3 thoughts on “The Blockchain Bubble will Pop, What Next?”

  1. Thanks for good posting again 🙂
    Do you have any plan for treating the fundamental research directions about AI(or ML) in this site?
    I wonder what AI discipline is actually, is the direction of AI research actually making AGI(well i can’t be sure you like this term or not)?

  2. A long time ago when I worked in materials science research, I noticed a number of aggressive projects that combined two not-very-well-understood advanced materials in an attempt to produce a new material that had none of the flaws and all of the advantages of the two parents.

    As a way to manage the risks associated with developing new materials — these were projects that were intended to produce practical solutions that could be incorporated into the next generation of military equipment — this is insane, but it is an effective way to get funding.

Leave a Reply

Your email address will not be published. Required fields are marked *