Terence Tao warned on November 4 that delegating thinking to AI erodes understanding of the answers
In a October 16 talk at UCLA, Tao said AI progress in mathematics is changing how humans solve problems, raising concern that solutions arrive without graspable reasoning.
Reported on November 4, Fields Medalist Terence Tao recently warned about growing human reliance on artificial intelligence. He is concerned that as people increasingly hand the process of thinking and problem solving to AI, humans are losing grasp of the problems themselves and of their solutions, and that this trend could soon produce serious consequences in the real world. The remarks came from Tao's talk at the Science x AI Summit held at the University of California, Los Angeles on October 16, whose full video titled Math 2.0 was recently posted by the Science and AI Research Foundation and whose clips circulated on social platforms.
Tao said we are delegating our thinking to AI, letting it solve problems for us, while at the same time losing a kind of guarantee, namely the ability to be sure that the solution AI gave is actually the one we wanted. He said he is genuinely worried we will soon see the consequences of this in the real world, adding that this is really a concern of his outside the field of mathematics. He noted that AI progress in mathematics is changing the traditional way humans solve problems: previously a mathematician attacking a hard problem needed long study and real understanding of the relevant theory, accumulating new knowledge along the way, whereas now AI can produce answers quickly and humans may not understand the reasoning behind them.