it would be maybe interesting to know how was Lem thinking about the actual algorithmic process the machine uses while writing this:
- purely symbolic with myriads of symbolic (if-else) clauses similar to expert systems, or
- sub symbolical ANN networks similar to current LLMs
i suspect it was the former in which case that he actually arrived at description of current large NNs is really striking I think