Neural networks were explored by a small group of people for about 80 years.

Judge as of January 2026. True: history of AI shows neural networks originated about 80 years earlier (McCulloch-Pitts 1943) and were pursued by a relatively small research community for most of that time until the deep learning breakthrough (around 2012). Mostly true: the timeline is roughly right (~70–85 years) and the field was small for most of it, with periods of wider interest (e.g., 1960s perceptrons, 1980s connectionism boom) as a minor caveat. Mixed: the origin date is roughly right but the field was large or mainstream for substantial periods, or explored by a small group for markedly fewer years. Mostly false: the small-group characterisation applies only to brief periods, or the time span is substantially wrong. False: neural networks were not explored for anything close to 80 years or were a large mainstream field throughout.

Set before the research began.

1. E1 History of artificial neural networks - Wikipedia Wikipedia · 18 August 2019

   “Despite the frequent claim that little research was conducted on ANNs in the 1970s and 1980s—the AAAI calling this period an "AI winter"[1]—there was ongoing development of ANNs, especially connected to the development of specialized computers for simulating these networks.”

2. E2 Deep Learning in Neural Networks: An Overview Technical Report IDSIA-03-14 / arXiv:1404.7828 v4 [cs.NE] (88 pages, 888 references) arXiv (IDSIA technical report, by Jürgen Schmidhuber) · 8 October 2014

   “Early NN architectures (McCulloch and Pitts, 1943) did not learn. The first ideas about UL were published a few years later (Hebb, 1949).”

3. E3 How a stubborn computer scientist accidentally launched the deep learning boom Ars Technica · 11 November 2024

   “Neural networks had delivered some impressive results in the late 1980s and early 1990s. But then progress stalled. By 2008, many researchers had moved on to mathematically elegant approaches such as support vector machines.”

4. E4 How a stubborn computer scientist accidentally launched the deep learning boom Ars Technica · 11 November 2024

   “But in 2012, a team from the University of Toronto trained a neural network on the ImageNet dataset, achieving unprecedented performance in image recognition.”

E2 dates the first neural network architectures to McCulloch and Pitts in 1943, about 83 years before the claim, which fits "80 years". E3 says neural networks produced notable results in the late 1980s and early 1990s, then stalled, and that by 2008 many researchers had moved to other methods. E4 places the breakthrough at 2012. E1 says neural network development continued through the 1970s and 1980s, despite a common claim that little research was done then, and E3 reports impressive results in the late 1980s and early 1990s; this points to periods of more activity, a minor caveat to the small-group description. The timeline is roughly right and the field was small for most of it, with some periods of wider activity, which matches "mostly true" under the criteria.