I don't think the work has become irrelevant at all. ML models are fine but they're really just big function approximators. In the context of Minsky's book they

I don't think the work has become irrelevant at all. ML models are fine but they're really just big function approximators. In the context of Minsky's book they're the sort of processes or agents which when put together and interacting could maybe constitute a generally intelligent system. Which is how they actually tend to be used in the real world already, as parts of more complex systems that interact or communicate.