In this Q&A podcast, Riley Tao interviews her father, Terence Tao, about the use of AI in mathematical research. Terence explains that AI is a powerful but currently unreliable tool, like a jet engine, and is best used in conjunction with human oversight. He details how he uses AI for literature reviews, coding, and exploring mathematical configurations, citing his work with Google DeepMind's Alpha Evolve as an example. He also addresses the issue of AI errors and the need for verification methods, comparing AI's output to a "fire hose of stewed water" that needs filtering. He expresses optimism about the future of math and science, where AI assists mathematicians and allows for broader public participation in solving complex problems, while also cautioning against over-reliance on benchmarks and advocating for real-world practicality and the exploration of alternative AI architectures.
Sign in to continue reading, translating and more.
Continue
