What is an AI superforecaster? How AI puts honest odds on the future
An AI superforecaster researches a question about the future and gives a calibrated probability. How it works, how it's scored, and where it stands in 2026.
Read the article
Blog
Notes on forecasting.
How honest probabilities are made, checked and scored, and what they say about the world’s biggest events.
Can AI predict the future? What the evidence says in 2026
AI can't see the future, but it now estimates the odds of real events close to the best human forecasters. What the 2026 research shows, and its limits.
Read the article
The Brier score, explained with examples
The Brier score measures how close a probability was to what happened. Lower is better; always saying 50% scores 0.25. Formula, examples and good scores.
Read the article
What is a calibrated forecast? Why 70% should happen 7 times out of 10
A calibrated forecaster's 70% calls come true about 70% of the time. What calibration means, how to check it, and why people and AI are often overconfident.
Read the article
AI vs. superforecasters: who predicts better in 2026?
The best AI is now statistically tied with superforecasters on ForecastBench, while top humans still lead on Metaculus. What each test shows, and what's fair.
Read the article
How prediction markets work, and what their odds really mean
A prediction market contract pays $1 if an event happens, so 30¢ means about 30%. How prices form, how Kalshi and Polymarket work, and when odds mislead.
Read the article
The favorite-longshot bias: why unlikely events look likelier in markets
In betting and prediction markets, long shots are overpriced. On Kalshi, buyers of contracts under 10¢ lost over 60%. What the bias is and why it happens.
Read the article
Base rates: the first step in every good forecast
A base rate is how often similar things have happened before. Start there, then adjust for the evidence: the surest way to avoid overconfident forecasts.
Read the article