I’m not anti-AI like most of the people here on Lemmy, but I’m certainly not as bullish on AI R&D as the author. Current techniques are not sufficient to replace expert humans, and it’s unclear if they ever will be.
Right now I expect AI to grow to replace non-experts in most text-only jobs, but it will probably take a full redesign of the algorithms to go beyond that. Like, you can only get so far by doing the statistically correct thing - research progress happens by thinking outside the box.