Centaur AI Model Passed Psychology Tests Without Following New Instructions
TL;DR: A 2026 study in National Science Open argued that Centaur, an AI model promoted as a simulator of human cognition, could reproduce psychology-test answers while failing a simple instruction-understanding check. Key Findings Centaur was tested as a cognition simulator: the earlier model reportedly performed well across 160 psychology tasks, including decision-making and executive-control experiments. …
