For decades, people have been saying that the prediction market has the potential to become economically important, yet it remains unimportant. I would not be surprised if it becomes important over the next 4 years thanks to broadly-available AI technology.

Let's define "economically important" as a state of affairs in which there continues to be at least $50 billion riding on predictions at every instant in time.

First of all, AI tech might make prediction markets better by helping with market-making and arbitrage. Second, a sufficiently robust prediction market might turn out to be the most cost-effective way for an owner of an AI service like ChatGPT to make certain kinds of improvements to the AI service with the result that owners of AI services become a major source of investment in and of revenue for the prediction markets.