Respect the P vs. NP Problem (blog.computationalcomplexity.org)

The ongoing debate surrounding the P vs. NP problem remains unresolved, despite recent advancements in AI, which have sparked interest in the potential for AI-generated proofs. The article emphasizes the distinction between viewing this problem as a formal conjecture and its broader implications in computation—primarily that something verifiable must also be computable efficiently. While AI has made strides in mathematical proofs, such as disapproving the Erdős unit distance problem, the author argues that it is unlikely we will see significant progress on P vs. NP in the near future, whether by human or AI efforts. The complexity surrounding the P vs. NP problem cannot be overstated; proving P ≠ NP requires demonstrating the failure of all conceivable polynomial-time solutions to NP-complete problems. Current methodologies like relativization and diagonalization have not yielded a viable path forward, and attempts through polynomial-time understanding have faced significant barriers. As a result, the article advises against formal approaches using tools like Lean for this intricate problem, suggesting instead that future breakthroughs will likely stem from intuitive insights rather than strict formalism.