Terence Tao criticizes the way AI proves mathematics: automated, predatory mining
In a recent interview, Terence Tao criticized the way AI proves mathematics, describing the chaos surrounding the Navier-Stokes equations as a kind of automated, predatory exploitation and mining. He believes mathematicians want AI to serve as an on-call supercalculator, not as a competitor fighting over results; the value of humans solving difficult problems lies in refining conceptual systems and theoretical tools through long-term derivation, whereas AI, relying on computing power and black-box logic to directly take the answer, leaves behind theoretical ruins that are hard to understand and impossible to build on. He emphasized that a correct answer does not equal scientific progress.
Why read this
AI score 52
Terence Tao's criticism of AI proving mathematics points to an often-overlooked problem: getting the right answer does not equal advancing the discipline.
Topics
Body · AI translation
Terence Tao, widely recognized as the leading figure in mathematics today, has issued an extremely harsh public criticism in a recent interview of the way AI proves mathematics—and this has nothing at all to do with whether the proofs are actually correct.
Many laypeople, when they see AI conquer a difficult mathematical problem, care only about the final cheer of "proof complete."
They assume that as long as the machine can spit out a correct solution step, it is yet another great victory of technology over human intelligence,
and the media and onlookers rush in, enthroning results forcibly derived by computing power, believing that yet another corner of a Millennium Prize Problem has been breached.
But Tao directly ripped away the fig leaf of this revelry.
Speaking about the recent chaos surrounding the Navier-Stokes equations, he used an extremely striking characterization: this is a kind of automated, predatory, extractive mining.
What mathematicians truly long for is for AI to serve as a tireless, always-on-call supercalculator, not a competitor who plays dirty and comes to fight humans over the fruits of their work.
Behind this lies a piece of scientific common sense that ordinary people completely fail to understand.
When human mathematicians solve a difficult problem, it is never for the sake of that one isolated answer,
but rather, over decades or even centuries of derivation, they gradually polish out entirely new conceptual systems, build rigorous theoretical tools, and along the way nourish and drive the whole discipline's unending development.
Today's AI, by contrast, is like a heavy bulldozer crashing into a primeval forest: relying on brute computing power and black-box logic, it violently grabs all the fruit in sight, leaving behind a wasteland of theory that humans can neither understand nor continue to cultivate.
Although humanity might, over many years, barely manage to understand that machine proof, this kind of advancement—killing the goose that lays the golden eggs—can be called extremely suboptimal, even deeply irresponsible.
Here is an extremely cold, counterintuitive truth:
A correct answer does not equal scientific progress. If you obtain the correct endpoint but blow up every theoretical ladder along the way to that endpoint,
this predatory computing power that cares only about immediate delivery is, in essence, personally overdrafting the future vitality of this discipline.
Stop blindly worshipping the ultimate answers spat out by black boxes.
The true dignity of humanity's exploration of the unknown has never been about stumbling by sheer luck onto a result, but about our need to know clearly and completely how we actually walked there, step by step. https://t.co/K4RoZAU2Ul