Stefan Feuerriegel Featured in Tagesschau

Tagesschau recently featured MCML PI Stefan Feuerriegel and the LLM SoccerArena project, developed at LMU Munich in collaboration with researchers from the University of Cologne and Paderborn University. The platform uses the FIFA World Cup as a real-world benchmark to evaluate how well large language models can predict soccer matches and tournament outcomes.

The article explains how models such as GPT, Claude, Gemini, DeepSeek, Qwen, Grok, and Mistral compete on an online leaderboard, with predictions scored according to their accuracy. It also highlights how differences between model predictions can offer insights into AI reasoning and decision-making.

MCML Junior Member Jonas Schweisthal also comments on how the models weigh information such as team form, injuries, rankings, and public opinion. According to Stefan Feuerriegel, the project can help researchers better understand AI-based decision-making beyond sports, including in business contexts.