Can AI predict the future? What the evidence says in 2026

Short answer: AI cannot know the future, but it can now estimate the odds of real-world events close to the level of the best human forecasters. Not with certainty, and not on every kind of question. But the gap between "AI that guesses" and "AI that forecasts well" has closed faster than almost anyone expected.

Here is what the research actually shows, where the limits are, and how to judge any claim that an AI "predicts the future".

Prediction versus forecasting

The phrase "predict the future" suggests a yes or no answer: this will happen. Good forecasting never works that way. It gives a probability: this is 30% likely. That difference matters, because a probability can be checked.

If a forecaster says 30% on a hundred different questions, about thirty of those events should happen. That is called calibration, and it is the only honest way to judge anyone, human or machine, who claims to see ahead.

So the real question is not "can AI predict the future?" but: can AI give probabilities that are both calibrated and informative?

What the research shows

2023: out of the box, models were close to chance

When researchers first tested large language models on questions that resolved after their training cut-off, most were barely better than guessing. Asked cold, without research tools, the models had no reliable way to reason about events they had never seen.

2024: add research, and AI approaches the human crowd

A team at UC Berkeley built a system that searched for news, reasoned over it and combined several forecasts. On questions from competitive forecasting platforms, it came close to the aggregated forecasts of the human crowd, and in some settings beat it (Halawi et al., 2024).

The same year, a study with 991 people found that forecasters who could consult an AI assistant were 24–28% more accurate than a control group (Schoenegger et al.). AI was already useful as a forecasting partner.

2026: near the best humans, with a caveat

By mid-2026, the Forecasting Research Institute reported that the top AI systems on its ForecastBench benchmark were statistically indistinguishable from superforecasters (FRI). Good Judgment pushed back that the human numbers were collected two years earlier on different questions (Good Judgment), and on Metaculus, top human forecasters still lead the best bots head to head (Metaculus).

We cover that debate in AI vs. superforecasters: who predicts better?.

Why AI got better so fast

Three changes did most of the work:

1. Research, not recall. Modern systems search for current evidence instead of relying on what they memorised during training.
2. Many views, combined. Several independent forecasts are averaged. Independent errors cancel, which is why a crowd often beats an individual.
3. Calibration against a track record. Raw model probabilities tend to be overconfident. Adjusting them against past results makes a 70% mean 70%.

Together these turn a language model into what is now called an AI superforecaster.

Where AI forecasting still struggles

• Thin or brand-new situations. With no history to compare against, there is no base rate to start from, for humans or machines.
• Leakage in tests. If a model saw the answer during training, a "forecast" is really memory. Only benchmarks built on truly future questions, like ForecastBench, avoid this.
• Question quality. A forecast is only as good as the question. Vague questions, or markets that resolve on technicalities, produce misleading scores.
• Overconfidence. Language models sound sure. Without calibration, their 90% can behave like 70%.
• Genuinely random events. Some things are close to a coin flip. The best possible forecast there is honest uncertainty, not a confident call.

How to judge a claim that "AI predicts the future"

Ask four questions:

1. Were the questions truly in the future when the forecasts were made?
2. Is it scored with a proper rule such as the Brier score, across many questions?
3. Is it compared with a real baseline: the crowd, the market, or top human forecasters?
4. Are the misses published, not only the hits?

If any answer is no, treat the claim as marketing.

Key takeaways

• AI cannot know the future, but it can estimate the odds of real events, and now does so close to the level of top human forecasters.
• The leap came from research tools, combining several forecasts, and calibration, not from bigger models alone.
• In 2026 the best systems match superforecasters on at least one benchmark; top humans still lead on others.
• Judge any AI forecaster by calibration and proper scores on truly future questions, with the misses included.

FAQ

Can AI predict elections or the stock market?

It can estimate probabilities for clear, checkable events such as an election result by a fixed date, and some systems do this well. Short-term market moves are much closer to random, and no forecaster, human or AI, should present them as predictable. Nothing here is investment advice.

Is AI better than humans at forecasting?

Better than the average person, clearly. Compared with the very best human forecasters, it is close, and the evidence is mixed: some benchmarks show parity, others still show top humans ahead.

How do I know an AI forecast is trustworthy?

Look for a published track record scored with a proper rule such as the Brier score, on questions that were in the future when the forecasts were made, including every miss.

Sources

• Halawi et al. (2024): Approaching Human-Level Forecasting with Language Models
• Schoenegger et al.: AI-Augmented Predictions: LLM Assistants Improve Human Forecasting Accuracy
• ForecastBench: About
• Forecasting Research Institute: AI models have likely reached parity with superforecasters on ForecastBench
• Good Judgment: The Hype and the Evidence about AI Forecasting
• Metaculus: FutureEval, Metaculus’s AI forecasting benchmark