AI agents are now designing and running their own experiments to settle scientific debates without any human help.
Automated systems are replacing the slow and biased process of human scientific debate. These AI agents use program synthesis to build competing theories and then test them against each other in digital simulations. This closed-loop process finds the most accurate explanation for data much faster than a team of researchers could. It removes the ego and tribalism that often stall progress in the cognitive sciences. Scientists can now spend their time analyzing final answers instead of arguing over experimental design.
Automated Adversarial Collaboration for Advancing Theory Building in the Cognitive Sciences
arXiv · 2604.25521
Cognitive science often evaluates theories through narrow paradigms and local model comparisons, limiting the integration of evidence across tasks and realizations. We introduce an automated adversarial collaboration framework for adjudicating among competing theories even when the candidate models and experiments must be discovered during the adjudication process. The system combines LLM-based theory agents, program synthesis, and information-theoretic experimental design in a closed loop. In a