In recent years Minsky lamented the 'success' of his book and had become respectful of recent gains in neural net technology.

Two elements: in his later years Minsky (a) expressed regret about the effect of 'Perceptrons' and (b) expressed respect for recent neural network advances. True: Minsky's own recorded statements (interviews, writings, talks) or independent accounts support both elements. Kurzweil's account alone is not sufficient. Mostly true: both elements are supported, but one only weakly or indirectly. Mixed: one element is supported and the other is not, or is contradicted. Mostly false: the support rests only on Kurzweil's account and Minsky's documented statements cut the other way on one element. False: Minsky's documented late statements show he remained dismissive of neural nets and did not regret the book.

Set before the research began.

1. E1 Artificial Intelligence Pioneer Marvin Minsky Dead at 88 - EE Times EE Times · 28 January 2016

   “However, he considered neural networks a dead-end study in the 1970s favoring AI, which could be controlled with clear and understandable “expert system” rules written by programmers and engineers. Neural networks could capture knowledge, he admitted, but their internal structure is just as decipherable as that of real brains. Finally, however, Minsky embraced neural networks due to limitations in the AI “expert systems” he so loved.”

2. E2 Marvin Minsky: The Brilliant Iconoclast ishistory (Minds & Machines series) · 3 May 2026

   “Minsky himself, in later years, expressed some regret about the impact of the Perceptrons book. He acknowledged that it had been read as a more definitive refutation of neural networks than he had intended”

3. E3 Marvin Minsky: The Brilliant Iconoclast ishistory (Minds & Machines series) · 3 May 2026

   “He was criticising them again in the 2010s, and it remains to be seen whether this second round of criticism will prove similarly premature.”

4. E4 Marvin Minsky, Ph.D. | Academy of Achievement Academy of Achievement

   “Meanwhile, Minsky revisited his consideration of neural nets and pursued the possibilities of parallel processing — the simultaneous handling of information by multiple agents, as it occurs in the brain and as it can occur in manmade systems.”

E2, a secondary profile, says Minsky expressed some regret in later years that Perceptrons was read as a more definitive refutation of neural networks than he intended, which supports the first element. On the second element the evidence conflicts. E1 says Minsky eventually embraced neural networks, but gives the limits of expert systems as the reason rather than respect for recent gains, and E4 says only that he revisited neural nets, with no date. E3 says he was criticising neural networks again in the 2010s, which cuts against respect for recent advances, so one element is supported and the other is not.