OpenAI has been making fast progress on maths problems using AI. Some of its systems can now solve questions that once took experts years. The Verge spoke to one of the mathematicians who advises the company.
Researchers in the field expect a new wave of AI maths results. Some worry these results could upend the way maths is done and taught. Others see a chance to speed up discovery.
OpenAI has faced criticism for moving quickly and not involving mathematicians enough.
π The Backstory
Maths is often used as a test of how smart an AI system is. Problems have clear right answers, so they are easy to check. In recent years AI models have done better and better on these tests.
Many mathematicians are unsure how to feel about the change. AI could help them prove hard ideas. But it also raises questions about credit, jobs and whether the work still counts as human discovery.
π― Why It Matters
AI that is good at maths can help solve real problems in science and medicine. But it also changes the work of people whose skills took years to learn.
OpenAI has been making fast progress on maths problems using AI. Some of its systems can now solve questions that once took experts years. The Verge spoke to one of the mathematicians who advises the company.
Researchers in the field expect a new wave of AI maths results. Some worry these results could upend the way maths is done and taught. Others see a chance to speed up discovery.
OpenAI has faced criticism for moving quickly and not involving mathematicians enough.
Maths is often used as a test of how smart an AI system is. Problems have clear right answers, so they are easy to check. In recent years AI models have done better and better on these tests.
Many mathematicians are unsure how to feel about the change. AI could help them prove hard ideas. But it also raises questions about credit, jobs and whether the work still counts as human discovery.
AI that is good at maths can help solve real problems in science and medicine. But it also changes the work of people whose skills took years to learn.