A Nobel Prize for Plagiarism (people.idsia.ch)

An opinion piece titled "A Nobel Prize for Plagiarism" argues that plagiarism among human AI researchers—distinct from legal disputes over training data—has gone largely unchecked and that a recent high-profile award celebrates work without properly crediting earlier discoveries. The essay surveys historical precedents it says were omitted: the 1925 Lenz–Ising recurrent idea and Shun‑Ichi Amari’s adaptive associative nets (1969–1972) preceding Hopfield’s 1982 model; early deep‑learning layer‑wise training by Ivakhnenko & Lapa (1965) and unsupervised pretraining work in 1991 predating celebrated 2006 methods; early attention architectures by Schmidhuber (1990–93); and antecedents to backpropagation (Linnainmaa 1970, Werbos 1982) and ReLU/neocognitron ideas (Fukushima 1969). The author also highlights citation errors (e.g., inflated Google Scholar counts) and points readers to a longer reference list documenting many alleged misattributions. The piece stresses why this matters: accurate attribution upholds scientific ethics, mentorship norms, and the integrity of prize decisions; unresolved miscrediting can distort citation networks, patent claims, and the historical record of technical progress. It urges corrective measures—errata, transparent timestamps, stricter bibliographic diligence, and even revocation of honors if plagiarism is proven—invoking ACM ethical guidelines as the standard for professional conduct.