Terence Tao: AI Solves Math Problems Faster Than They Can Be Replaced

This signals a structural shift in how frontier mathematical research is consumed, not just produced. If AI compresses the lifespan of open problems, the bottleneck moves from solving to problem generation and validation, which could redefine what 'hard' means in the field. For institutions and researchers reliant on sustained, incremental progress, the pace of obsolescence becomes a strategic concern. The open question is whether the rate of new, meaningful problem creation can keep pace with AI's flattening effect, or if research will pivot toward harder-to-formulate, applied or interdisciplinary challenges.