Cottom cuts to the heart of an issue that both AI doomers and AI boosters seem to take for granted: that an AI-dominated future isn’t inevitable.

Given the development plateau tech companies seem to be running into these days, a future built on large language models (LLMs) is far from a lock. (Consider, for example, the fact that after a very recent December update, ChatGPT couldn’t even generate an accurate alphabet poster for preschoolers.)
...
Cottom points to the historical record — for example, the fact that chattel slavery was at one time seen as a preordained fact of life, a myth spread by the wealthiest members of that bygone age.

I like this comparison