Terence Tao warns that outsourcing thinking to AI erodes understanding
Terence Tao warned that delegating problem-solving to AI can leave people unable to understand or evaluate the answers they receive. In his October 16 Science x AI Summit speech at UCLA, he said the risk becomes more complicated outside mathematics, citing power grids, air-traffic control, drug development, and chip design. The Science and AI Research Foundation later published the full “Math 2.0” talk on YouTube, and Tao also called for retaining human abilities to understand, verify, and judge AI-generated results.
Tao’s warning highlights a source-supported concern that AI systems may produce model-valid answers that users cannot fully assess or that may not match real-world needs in safety-critical fields.