Terence Tao and Daniel Tubbenhauer on Mathematics and AI
At 50:50 he makes an observation that AI doesn't have access to the process mathematicians go through in proving a difficult theorem. Indeed, student's don't even have access to it!
This institute was founded by Jim Simons with some of his ill-gotten gains, ... see Jane Street Capital.
Subscribe to Simons Foundation.
Daniel Tubbenhauer has started a whole series on this:
At the end of Terence Tao's talk (56:43) there is a question that connects with what Daniel says about the way mathematicians work. People who work in category theory often have the notion that there is a kind of natural and logical order in which theorems appear. When you don't work on them in that natural way then you can end up having to use convoluted and roundabout reasoning to get past areas where you have don't have the necessary knowledge. Then later people discover much more concise proofs which are direct and don't have these elaborate excursions around the darker areas of the intellectual landscape.
So I should make a prediction, shouldn't I? Just say that one day, hopefully within fifty years people will be able to prove the Four Colour Theorem, Fermat's Last Theorem and maybe even the Kepler conjecture in undergraduate textbooks.
Subscribe to VisualMath.
Comments
Post a Comment