[squidbeak]

The article's argument is pretty common - but it's looking for a precedent for a technological advance without a precedent. Cameras weren't capable of reasoning or thinking (or for the quibblers, predicting output well enough to be indiscernible from reasoning and thinking), and they weren't generalized.

I'm not a senior engineer and I trust those that tell me LLMs are currently incapable of high level coherent engineering. But at the same time, this is a rapidly developing tech whose progress only seems to be accelerating, and as long as it's gaining on every area of coding and planning, anyone arguing AI will never be equal or superior to human SWEs is wish thinking.

We've lived through all this before with the Go community, who used to be adamant a machine could never do better than a human - because of some ethereal semi-magical quality only existing in the skulls of bipeds and totally inaccessible to machines. SWE's Alphago Zero moment could be in the post.