Opaque Corpora Make the 'Stochastic Parrot' Theory Hard to Disprove

RexDouglass · x · 2026-08-10

Commenting on a recent AI-assisted mathematical proof, the original author points out that its core algorithms and predictions have a deep prior history in academic literature, rather than being entirely created from scratch by AI.

Meanwhile, another scholar uses this to discuss the 'stochastic parrot' theory. Because the training corpora of closed-source models are completely hidden, it is impossible to verify whether their innovations are genuine reasoning or simply regurgitation of existing knowledge. This opacity makes the theory currently impossible to disprove.

Original post →

More from AGI Musings

AGI Musings channel →