F. Chollet: all AI will converge to symbolic learning as the optimally efficient form

burny_tech · x · 2026-09-05

Citing F. Chollet: all AI will inevitably converge toward symbolic learning — modeling data by finding the shortest symbolic program that explains it — since that is the optimally efficient form of AI, though multiple evolutionary paths may lead there.

The quoted neurosymbolic programming survey notes the field's holy grail would be a general learning algorithm that efficiently generates arbitrary combinations of neural components and symbolic code. Such an algorithm hasn't been developed and seems hard, given the joint numerical and combinatorial optimization required; the community has only succeeded with specialized approaches for particular program-structure/component pairings.

Original post →

More from AGI Musings

AGI Musings channel →