Ben Recht revisits his 2019 AI reading list: the critiques predate GPT-2 and still hold

beenwrekt · x · 2026-11-19

Writing during "p(doom) week" timed to the Anthropic IPO, Ben Recht revisits the AI reading list he assembled in 2019, arguing that critics offer diagnostic premonitions, not forecasts — those who correctly call out inevitable outcomes of power imbalances get labeled haters, while "fulfilled" predictions get credited as sorcery. Ignoring the etiology makes predictions self-fulfilling.

Highlights:

All were written before GPT-2 — no LLMs existed, yet the stories were the same, and the critiques still hold. Recht's point: these texts are really about the people who make AI and their mythologies.

Original post →

More from AGI Musings

AGI Musings channel →