Current post-training paradigms can't diffuse every human skill, capping AI's job impact
menhguin · x · 2026-10-09
The author clarifies the assumption behind his claim that AI won't cause net job losses for a decade: under current training/post-training paradigms, essentially every human skill and workflow has to be intentionally post-trained into models, and there is no way to rapidly diffuse every possible skill in every proprietary context — which naturally caps AI's employment disruption.
More from AGI Musings
- David Patterson predicts SI and robots will rapidly replace jobs by 2029-30 — davidpattersonx · 2026-10-09
- Yacine: courage × intelligence has an optimum — don't be smart enough to think it through — yacineMTB · 2026-10-09
- Researcher slams sweeping claims on consciousness as failing undergrad-level rigor — dioscuri · 2026-10-09
- 2021 Was a Dead Cat Bounce — Only ChatGPT Saved Tech Investing From Ideas Drought — menhguin · 2026-10-09
- Dioscuri: most consciousness scientists back mainstream cognitive theories — dioscuri · 2026-10-09
- Yi Ma and Yann LeCun spar over automation's coming reshaping of mathematics — CSProfKGD · 2026-10-09