Experts' "AI Will Never Do X" Predictions Keep Getting Falsified
Glen_Garrett_Gayhart · reddit · 2026-11-22
A Reddit post argues that today's confident "AI will never be able to do X" claims echo the historical "flying machines which do not fly" skepticism.
- The pattern: an expert declares AI can't do X, and a headline the next day announces AI doing X for the first time.
- The author says the "stochastic parrot" hypothesis — that AI only remixes what it has heard — is being falsified at an alarming rate.
- The post is opinion-driven without a case list, but highlights a systematic gap between predictions and outcomes.
Related event: Crowd-sourced compilations mock 'AI will never' predictions(3 posts)→
More from AGI Musings
- If AI Does the Work, Who Gets to Learn? Lessons from Netscape to Anthropic's $81.3B IPO — shivsingh · 2026-11-23
- The great convergence: AI is turning every role and tool into the same builder — bendee983 · 2026-11-23
- WSJ: More twentysomethings skip entry-level jobs to become their own bosses — himanshustwts · 2026-11-23
- So8res: Anthropic fails its ideals as Dario dodges extinction-risk questions — trevposts · 2026-11-23
- Economist: AI ends the premium on technical grunt work — asking questions matters more now — _akpiper · 2026-11-22
- WaPo investigation: Amodei privately put 20-30% odds on AGI destroying humanity — GaryMarcus · 2026-11-22