AI predicts the future; it can now forecast stock market trends or the chart performance of a Shakira song just as well as—and soon better than—top specialists. It is a market that several startups are looking to develop.
By 2030, “the way we make decisions today will seem like the Stone Age to us,” claims Dan Schwarz, co-founder of the forecasting firm FutureSearch, who envisions AI significantly improving our strategic choices.
Fueled by vast amounts of data—some public, some purchased from vendors—AI forecasting models are already neck-and-neck with the best “superforecasters”: gifted individuals whose estimates far surpass those of the average person.
In early September, two AI systems took the top spots in the Metaculus Cup—a competition pitting human and software forecasting champions against one another—marking a historic first.
FutureSearch’s in-house wizard, engineer Tom Liptay, is aware that his job is not secure in the medium term, notes Dan Schwarz. “He knows that these [programs] are going to outperform him.”
By using artificial intelligence on the US prediction platform Kalshi, Preseen co-founder Theo Summer managed to turn $35 into $1.9 million in seven months.
The goal is not to guarantee a specific outcome but to determine the probability of an event occurring, while allowing for the cross-referencing of hypotheses linked to multiple interconnected events—such as fluctuations in oil prices and the war in the Middle East. “So far, hedge funds have been the quickest” to purchase the service, explains Veniamin Veselovsky, co-founder of Preseen, “because as soon as they think they can make money from something, they can position themselves quickly.”
Algorithms have already been shaping Wall Street for over a decade without human intervention, but the emergence of AI-driven forecasting is expected to further accelerate the automation of financial markets.
“I think it’s already having a small impact, as we have clients using it to execute trades,” says Toby Shevlane, head of Mantic, another player in this burgeoning industry.
AI is also taking the now-famous prediction platforms Kalshi and Polymarket—used by millions of Americans—by storm.
“Time is running out if you want to trade on these markets” and hope to turn a profit, warns Veniamin Veselovsky, because “they will soon be dominated by AI forecasting systems.”
Monitoring risks
Moreover, these products are intended—at least in the medium term—for professional clients and corporations, remaining inaccessible to individual investors due to their high cost.
The most advanced models do more than just predict a specific event; they can formulate their own questions about the future and then assign probabilities to various outcomes, notes Toby Shevlane.
The reach of AI forecasting is beginning to extend beyond finance. Mantic provides a major US corporation with probable scenarios regarding regulatory changes in its sector and has also worked for the British government. As for companies, “they are interested in risks and how to monitor them,” using probabilities provided by AI, says Toby Shevlane.
Preseen prefers to offer its services rather than setting up its own investment vehicle based on its models, “because the market of decision-makers is much larger (in terms of potential revenue) than the money a fund could generate,” argues Veniamin Veselovsky.
Dan Schwarz points to applications in medical research, particularly regarding funding decisions between various avenues of exploration that could lead to a treatment.
“Policy choices by governments, logistics decisions by companies, or risks to be covered by insurers—that is where the value for society and the revenue for AI forecasting lie,” suggests Veniamin Veselovsky.