AI Reshapes Math Research via Automated Proofs

Professors Michael Kearns and Aaron Roth from the University of Pennsylvania report a major shift in scientific methodology ^[1]. AI systems are transitioning from passive assistants to active collaborators in mathematical research ^[1][3]. These tools now generate fully structured proofs from conceptual outlines provided by human researchers ^[1][2]. Known as agentic AI, these systems handle the technical rigor of constructing complex arguments ^[2]. Researchers provide high-level ideas, while the software executes the formal validation ^[1]. This collaboration allows scientists to focus on broader theoretical questions instead of manual proof steps ^[2][3].