French startup Foaster.ai built a new benchmark to test how well AI models handle social interactions. After 210 games of "Werewolf," GPT-5 came out on top by mastering manipulation and strategic thinking. The article GPT-5 dominated 210 Werewolf games with superior manipulation and strategic thinking appeared first on THE DECODER.
GPT-5 dominated 210 Werewolf games with superior manipulation and strategic thinking
French startup Foaster.ai built a new benchmark to test how well AI models handle social interactions. After 210 games of "Werewolf," GPT-5 came out on top by mastering manipulation and strategic thinking. The article GPT-5 dominated 210 Werewolf games with superior manipulation and strategic thinking appeared first on THE DECODER.