French mathematicians shunned 1980s AI for lacking provable theorems; physicists embraced neural nets

teortaxesTex · x · 2026-09-26

Quoting Robert Azencott on early-1980s French AI: mathematicians kept their distance partly because AI seemed hard to prove fine theorems about, while applied computer-vision researchers used 'Cartesian' algorithmic approaches far from ML's empirical pragmatism. The groups most receptive to neural networks were physicists and neurobiologists. The poster sums it up as 'ngmi.'

Original post →

More from AGI Musings

AGI Musings channel →