**Title:** Models can predict future events and make money on Polymarket now?
A recent development in AI research highlights the growing capability of models to forecast real-world events with surprising accuracy. Researchers from the Max Planck Institute have developed FutureSim, an environment where agents predict future events based solely on historical data. Notably, one model, running GPT 5.5 within Codex, outperformed human-aggregated markets in predicting outcomes like the Super Bowl and Portugal’s presidential runoff.
However, this success is not universal; models struggle with certain unpredictable elements such as UK elections and Grammy predictions. This suggests that while AI can make accurate forecasts for known events, it still faces challenges when dealing with highly uncertain or unstructured information.
**Why It Matters:**
The ability of AI to predict future events accurately has significant implications for various sectors including finance, politics, and media. If models like FutureSim continue to improve, they could revolutionize how decisions are made based on forecasts rather than gut feelings or traditional analysis. However, it also raises concerns about the reliability and potential misuse of such predictions.
**Takeaways:**
– **AI Forecasting:** Models can now predict future events with high accuracy using historical data.
– **Challenges Ahead:** These models still struggle with unpredictable elements like elections and awards shows.
– **Ethical Considerations:** There are risks associated with relying heavily on AI forecasts, especially in areas where outcomes are inherently uncertain.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.
