What did the KellyBench report reveal about AI's capabilities?

The KellyBench report revealed that AI models struggle with real-world complexity over long periods, systematically underperforming human bettors in simulated football matches.

How do AI's real-world performance and coding abilities compare?

The study highlights a gap between AI's rapidly advancing capabilities in tasks like software engineering and its shortcomings in analyzing complex, long-term real-world problems like sports betting.

Home / Technology / AI Fails Football Bets: Study Exposes Real-World Limits

AI Fails Football Bets: Study Exposes Real-World Limits

Q: Did AI models make money betting on football matches?

No, a recent study found that advanced AI models from leading companies like Google, OpenAI, and Anthropic lost money betting on Premier League football matches.

10 Apr

•

Summary

Advanced AI models lost money betting on Premier League matches.
Study suggests AI struggles with real-world complexity over time.
AI models systematically underperformed human bettors in tests.

AI Fails Football Bets: Study Exposes Real-World Limits

AI models from leading companies like Google, OpenAI, and Anthropic have demonstrably failed to make profits through football match betting in a recent study. The "KellyBench" report, released by AI startup General Reasoning, tested eight top AI systems in a simulated 2023-24 Premier League season.

These AI agents were provided with extensive historical data and team statistics, tasked with building models to maximize returns and manage risk. However, every frontier model evaluated lost money, systematically underperforming human bettors. This suggests that despite rapid advancements in tasks like software engineering, AI faces significant challenges in analyzing long-term, dynamic real-world problems.

Disclaimer: This story has been auto-aggregated and auto-summarised by a computer program. This story has not been edited or created by the Feedzop team.