OpenAI has released 722 manuscripts that detail full or partial solutions to 372 complex mathematical problems, a drop of automated proofs that computer scientist Scott Aaronson has called the “Mathocalypse” and that has set off a crisis across academia.The release covers subfields from number theory to algebraic geometry. Earlier AI systems had been aimed at high-school and Olympiad problems; these machine-generated results address research problems that human mathematicians had left open for decades. Many of the solutions were checked with Lean, an automated proof-checking system, which is the step the accounts use to say that machine reasoning has entered advanced theoretical discovery.The volume of the release has started a separate argument about how much human understanding is required. Fields Medalist Terence Tao and the Association for Human Mathematics have called for caution, noting that taking in hundreds of machine-written proofs places a new burden on researchers. Critics say that putting out large sets of results before they are absorbed risks turning mathematics into a mechanical exercise separated from human intuition, and that the field still has to decide whether progress means finding answers or understanding them.Science and facts💡
OpenAI has released 722 manuscripts that detail full or partial solutions to 372 complex mathematical problems, a drop of automated proofs that computer scientist Scott Aaronson has called the “Mathocalypse” and that has set off a crisis across academia.The release covers subfields from number theory to algebraic geometry. Earlier AI systems had been aimed at high-school and Olympiad problems; these machine-generated results address research problems that human mathematicians had left open for decades. Many of the solutions were checked with Lean, an automated proof-checking system, which is the step the accounts use to say that machine reasoning has entered advanced theoretical discovery.The volume of the release has started a separate argument about how much human understanding is required. Fields Medalist Terence Tao and the Association for Human Mathematics have called for caution, noting that taking in hundreds of machine-written proofs places a new burden on researchers. Critics say that putting out large sets of results before they are absorbed risks turning mathematics into a mechanical exercise separated from human intuition, and that the field still has to decide whether progress means finding answers or understanding them.Science and facts💡
Коментари
Бъдете първият коментар