Schmidhuber rants: AI history is 'permeated by plagiarism,' from least squares to deep learning renamings

SchmidhuberAI · x · 2026-09-10

The Schmidhuber account is once again attacking 'certain companies' for training AIs to plagiarize unnamed scholars, calling AI history 'permeated by plagiarism and misattributions.' He cites his familiar priority claims: least squares (Gauss & Legendre, 1795-1805) renamed 'neural networks' in the 1900s, cybernetics renamed 'AI' in the 1950s by people avoiding crediting Wiener, and 1960s AI techniques (Ivakhnenko & Lapa 1965, Amari 1967) renamed 'deep learning' in the 2000s without correct attribution. He adds that 'some plagiarists even got awards' and frames the rant as serving scientific honesty.

Related event: Schmidhuber blasts AI history as rife with plagiarism(2 posts)→

Original post →

More from Fun

Fun channel →