Has the artificial intelligence bubble burst? The question has been on the lips of many this week as they have watched the stock market slide, with billions wiped off the value of technology companies that had boomed since the arrival of large language models. It was the economist Hyman Minsky who thought Wall Street encouraged too much risk and gulled investors into parting with cash for all the wrong reasons. The result was ruinous boom-and-bust cycles. His model began with “displacement”, which is when investors get excited about something they don’t fully understand – something like AI.
Last year, OpenAI’s ChatGPT chatbot got to 100 million users in two months. What follows displacement, according to Minsky, is boom, euphoria, profit-taking and panic. There’s been a boom. It is reported that more than 200 AI startups globally are unicorns – meaning they’re valued at $1bn or more.
There’s been plenty of euphoria too. Earlier this year, a report by the Wall Street Journal claimed that the CEO of OpenAI, Sam Altman, was in talks with investors to raise between $5tn and 7tn to reshape the global chip industry and thereby ensure the supply of the necessary computing power. The assumption here is that all that is needed to approach lifelike sentience is to increase computing power and make available the data to crunch. There are surely too many expectations for that bet to be a sure thing.
We could very well be in the stage of profit-taking. This is when there is a Wile E Coyote moment for stocks. When the cartoon character has run off a cliff, he looks down and realises that he’s standing on thin air – and plunges. With markets, investors suddenly clock that prices can’t just keep rising – and need to pull out their cash, causing share prices to crash. That’s been happening to semiconductor stocks, particularly Nvidia, which is the only company currently capable of producing the graphics processing unit chips that power AI.
Whether we have reached the stage of Minskyian panic is moot. The bubble bursts as quickly as investors can flee the markets. Amazon’s share price tumbled in part because analysts looked at its earnings and saw that the company spent lavishly on AI without much to show for it. Some venture capitalists think the party hasn’t even begun. Yet so far the public haven’t found much use for the technology. In a poll for the Reuters Institute, 29% of people in the UK said they had used ChatGPT, but only 2% used it daily. Most people who tried the chatbot, wrote the analyst Benedict Evans, “didn’t see how it was useful”.
Those concerns might bother the markets, but humanity has larger things to worry about. If AI’s development is not managed carefully, many argue, humanity risks engineering its own extinction. Politicians say the technology is advancing at such a rapid pace that there must be protections for the public. Notably, California is considering a law that would, in the event of misuse of an AI model, hold liable not only the party responsible but the original developer of that model. That pits the state’s key industry against many of its lawmakers. If anything ought to bring the AI juggernaut to a halt, then surely societal considerations are better than financial speculation.
