Ted Chiang makes a couple of mistakes. First, we know that human intelligence is a bell curve, so genetically additive. We have a good idea of how to increase human intelligence quite easily with technologies like CRISPR. We simply need more comprehensive GWAS. Second, DSLs can be functionally recreated by an API in a general-purpose language, so the point is moot. They are just pointless babble. Third, humans don’t have general intelligence but heuristics. Don’t overestimate humans. At the same time, don’t overestimate the present state of A.I. or neurology because they’re currently unimpressive.
Artificial Intelligence or Natural Stupidity?