François CholletVerified account

The specialization of human intelligence, perhaps paradoxically, is also the source of its generalization power.

Slightly off topic, there's good reason to be suspicious of "generalists" who are not also "specialists" in something. I don't think one can generalize usefully unless one already has access to a generalizable "structure" in which new knowledge can be rooted; sorted, interpreted.

*off topic, in the same vein. I tend to think that deeper the specialization (as long as the object of specialization could afford a powerful enough meta-language), stronger the generalizability to areas outside the specialization.