Will machines eat mathematics ? | Kevin Buzzard | TEDxImperialCollege
Will computers be able to do mathematical research, make conjectures, and prove theorems ? Kevin Buzzard gives an overview of the current state of the art of two tools: large language models and interactive theorem provers. Neither of them alone is anywhere close to being able to replace human mathematical researchers right now, but perhaps in the future, a combination of them will be able to make some progress. Kevin Buzzard got his Ph.D. from the University of Cambridge in 1995 and settled in London after staying in Berkeley, Harvard, and the Princeton IAS. He became a professor of pure mathematics at Imperial College London. This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at www.ted.com/tedx
Пікірлер: 4
Focusing heavily on large language models and mathematical theorem provers on the backend of a system that centralized mathematical content (and ideally automatically converts it into formalized math) might be one of the biggest steps humans will have thus taken, and his shirt, is why I love the mathematical community more than any other community. Mathematicians are the most humble. Not big headed like tech people or cocky like physicists, yet consistently the backbone to all major human revolutions.
I wish his presentation would be longer. It was to short for such an exciting topic.
He has very wrong notion of mathematics , He says relation between Automorphic Forms and Discrete Objects is Langlands Program but that is not Langlands Program is relation between Number Theoretic objects and Galois Representations further to Motives or Geometric Objects . For learning more about Langlands Program see the videos of Robert Langlands Abel Prize lectures.
@mindsetnovice
19 күн бұрын
For some reason I remember Kevin Buzzard's summer school "Automorphic Forms and the Langlands Program" at MSRI in 2017 covering the relation of number theoretic objects such as Galois representations not just with geometric things like Shimura varieties (motives) but with automorphic representations as well. The analytic ("automorphic") side of the theory was sort of the main area in which Langlands was an expert. Buzzard giving a watered-down version of Langlands here to the viewers of this TED talk probably doesn't represent any error in his "notion of mathematics." Although I'm sure that any videos of lectures given by Langlands are worth watching anyway.