Just read the last chapter: The authors appear dead set on a highly theoretical project of proving "learning theorems". They say that they've successfuly proved such theorems for a very weak class of perceptrons. They then insinuate various obstacles to proving such theorems for "Gamba perceptrons" (closer to what people use today in things like GPT) on the grounds of "vacuous generality". But the point they miss, is that the algorithm is more important than the type of automaton, and there's no reason to expect that the capabilities of a learning algorithm can be characterised well by a theorem. The actual direction of Machine Learning has been far more empirical than the one they set out.
Those are all the hallmarks of a rumour of the kind that spread on the internet, based on nothing else than half-digested information and a thirst for controversies. It should be noted that Minsky, himself, was a connectionist who worked on neural networks, as well as other approaches to AI. For example:
https://en.wikipedia.org/wiki/Stochastic_neural_analog_reinf...
I don't think most people who repeat those claims about Perceptrons have even read the book and really know what it says, other than from second- or third-hand sources (i.e. something someone once posted on twitter).
The entire book, in pdf format, can be downloaded from here:
https://ieeexplore.ieee.org/book/8076704
It's a whole book and not a blog post so it takes some reading. Alternatively, there is the introduction to the 1988 reissue with a short foreword by Leon Botu, focusing on the history of the book and its repercussions here:
https://leon.bottou.org/publications/pdf/perceptrons-2017.pd...
And a review of the book with more technical information here:
https://www.sciencedirect.com/science/article/pii/S001999587...