A recent study shows that advanced AI models perform poorly when betting on soccer matches and often incur losses, even though they have access to extensive historical data. Systems from major developers such as Google, OpenAI, Anthropic, and xAI were tested in a simulated Premier League season but struggled with real-world complexity and long-term decision-making.
Many models were unable to adapt their strategies, and some even went bankrupt. xAI’s Grok performed particularly poorly compared to the others. The results highlight a key limitation of current AI: while it is effective in controlled tasks, it still struggles in dynamic, unpredictable environments that require consistent reasoning and strategic adjustment.
