> The future of mathematics — “Math 2.0” — will require both expanding the research frontier, while simultaneously decentering the traditional role of problem solving.

Seems to be the crux of the argument, but "use your imagination" isn't a great thing to tell people who are looking at degree irrelevancy, concerned about getting tenure or a research position. How do we measure if someone is a good mathematician or not, if they are one of the sanctioned few who get access to the biggest AIs?

How do we measure if someone is a good mathematician or not?

Letters of recommendation from trusted colleagues have always been essential for evaluating candidates. Hopefully, human recommendations will remain a strong, faithful signal as the utility of other metrics rapidly deteriorate.

This consolidation of power is precisely why there is a need for open model development, and exactly why frontier labs have been lobbying hard to abolish them.