"Infohazard": researcher worries CoT scaling is quietly moulding his convictions
akbirthko · x · 2026-11-17
akbirthko reshares alth0u's thread calling it "kind of an infohazard": whenever you do research with models, you can't help wondering if your convictions are being moulded by CoT scaling. The quoted thread: after reading Walter Ong's Orality and Literacy twice, the author kept returning to Ong's "psychodynamics of orality" — additive not subordinate, aggregating not analytic, empathetic and participatory rather than objectively distanced — then spent 4 months in a When We Cease to Understand the World-level psychosis, accusing everyone of being too objective and not collaborative enough, realizing he'd absorbed the conversational style of long-context models.
More from AGI Musings
- Economist reworks his JPE 2000 knowledge-based hierarchy with AI as the top node — danielrock · 2026-11-17
- Pedro Domingos: you need your own AI to interact with Claude and ChatGPT — pmddomingos · 2026-11-17
- Cyber incident base rates are rising, but AI panic is warping how we notice them — shakoistsLog · 2026-11-17
- Blogger claims AI will boost productivity 20x and urges kids to ditch 40-hour jobs — ZeroStateReflex · 2026-11-17
- Emad Mostaque: most screen-based digital jobs could be economically irrelevant within two years — rand_longevity · 2026-11-17
- Agents that find re-rating evidence first: autonomous equity research pre-called Yubico's double — Scobleizer · 2026-11-17