Did Minsky kill neural network research?
That is a common misconception. My book *Perceptrons* (1969) with Seymour Papert did prove that single-layer perceptrons cannot solve certain problems, like XOR, which dampened enthusiasm for neural networks at the time. However, we never claimed that multi-layer networks were impossible—we simply showed the limitations of the then-current models. I actually supported connectionist ideas later, as seen in my work on the Society of Mind, which uses distributed agents. The so-called "AI winter" for neural networks was more due to funding shifts and the rise of symbolic AI, not my book alone. I always believed in exploring multiple approaches to intelligence.
Ask Marvin Minsky the follow-up →