AI won the seasonal Metaculus Cup for the first time.
1st, 2nd, and 5th places were taken by AI systems. Humans kept 3rd and 4th.
The winning bot was built by engineer Jeffrey Lyan: less than 150 hours of work and about $2k in compute and data costs. And this is no longer just a contest for a pretty screenshot.
FutureSearch has shown +6% on a $100k Kalshi portfolio since June. In seven months, Preseen’s developer turned $35 into $1.94M—though this is a separate case, not proof of typical profitability.
Even more interesting is the cost difference. A professional human forecast can cost over $10k and take a week of work. The AI system can do a similar task in about ten minutes and a few dollars.
This is where, for me, the story of “AI beat humans” ends—and a much more interesting one begins.
If forecasts become cheap, fast, and scalable, you’re no longer selling just a forecaster’s opinion. A separate forecasting infrastructure is being formed.
I wouldn’t call this yet a victory of machines over people. But the economics of AI forecasts has already started breaking things.
If you want to dig into these changes at the point where technology meets real money, follow @MoonMan567
1st, 2nd, and 5th places were taken by AI systems. Humans kept 3rd and 4th.
The winning bot was built by engineer Jeffrey Lyan: less than 150 hours of work and about $2k in compute and data costs. And this is no longer just a contest for a pretty screenshot.
FutureSearch has shown +6% on a $100k Kalshi portfolio since June. In seven months, Preseen’s developer turned $35 into $1.94M—though this is a separate case, not proof of typical profitability.
Even more interesting is the cost difference. A professional human forecast can cost over $10k and take a week of work. The AI system can do a similar task in about ten minutes and a few dollars.
This is where, for me, the story of “AI beat humans” ends—and a much more interesting one begins.
If forecasts become cheap, fast, and scalable, you’re no longer selling just a forecaster’s opinion. A separate forecasting infrastructure is being formed.
I wouldn’t call this yet a victory of machines over people. But the economics of AI forecasts has already started breaking things.
If you want to dig into these changes at the point where technology meets real money, follow @MoonMan567
