>>45 I'd say topology is the hardest. and quantum physics has the most potential

>>45 AI and machine learning have been accelerating in recent years.. and it still has the most potential for progress. >>47 I'd say topology is the hardest as well.

idk about hardest I'd say most potential is computational complexity theory, since it's one of the newest fields (and thus there's more 'low-hanging fruit' as it were) and results are already having implications in other unrelated fields (see the MIP* = RE proof for an example).