Startups Are Betting on AI to Predict the Future

By 2030, "we're going to feel that what we were doing for our decision-making was complete Stone Age thinking," Dan Schwarz, cofounder of the forecasting company FutureSearch, said.

AI forecasting models are already on par with the best "superforecasters" — gifted individuals whose estimates are markedly superior to those of ordinary people — thanks to troves of data acquired from both public and private sources.

In early September, two AI models made history when they took the top spots in the Metaculus Cup, a competition among prediction champions, both human and nonhuman.

FutureSearch's in-house human wizard, superforecaster Tom Liptay, is aware that "he's not secure in his job" in the medium term, Schwarz said. "He knows that these things are going to surpass him."

Theo Summer, who cofounded Preseen with Veniamin Veselovsky, used AI tools to turn $35 into $1.9 million in just seven months on the American prediction-making site Kalshi.

The idea is not to guarantee a result but to define a probability of occurrence, with the ability to cross-check assumptions related to several interconnected events — for example, changes in the price of oil and the war in the Middle East.

"Right now, the quickest buyers are normally the hedge funds, just because if they think they can make money off something, they can move pretty fast," Veselovsky said.

Algorithms have shaped Wall Street for more than a decade, but the rise of AI forecasting is likely to further increase the automation of financial markets.

"I think it's probably already having a small impact because we already have customers who use it when making trading decisions," said Toby Shevlane, the cofounder of another forecasting startup called Mantic.

Now-famous prediction markets Kalshi and Polymarket, which are used by millions of Americans, are also becoming inundated by AI use.

It might soon be too late for retail investors, who don't have access to expensive AI tools, to profit on those types of consumer-oriented trading sites because "they will soon just be dominated by AI forecasting systems," Veselovsky said.

The most advanced models no longer just predict events, either.

They can also formulate their own questions about the future and then assign probabilities based on the different possible outcomes, Shevlane said.

Their sphere of influence is beginning to extend beyond finance, as well, with potential applications in government policy, corporate decision-making, scientific research, and insurance risk assessments.

Mantic provides a large American company with likely scenarios of regulatory developments in its sector and has also worked for the British government. 

Companies "are interested in risks and how to monitor them," Shevlane said.

Meanwhile, other customers might be more interested in medical research, particularly to help make decisions around which types of treatments would be best to fund, Schwarz said.

Widespread access to these types of tools can come with costs, though.

"I think that good can be derived from this," Veselovsky said, but "I'm also a little bit scared because, you know, maybe Russia or China has better systems."