Dean Ball: AI Outputs Are Distinct, Undermining 'AI Threat to Democracy' Models
deanwball · x · 2026-08-09
Dean Ball points out that many 'AI and democracy' threat models silently assume AI outputs will be indistinguishable from human writing. This hasn't proven true; AI outputs have actually become more distinct from human writing since GPT-3.5, even beyond just being identifiable as 'slop'.
Related event: AI Content Not Indistinguishable, Threat to Democracy Premise Questioned(2 posts)→
More from AGI Musings
- AI Labs Lost the Societal Narrative, Even the Texas GOP Targets Datacenters — mattwbaker · 2026-08-09
- Pausing AI development would mean more inference compute, so people should favor it — kipperrii · 2026-08-09
- The AI Productivity Gap: Super-Savvy Users Yield 100x Power of Teams — bindureddy · 2026-08-09
- AI Makes Intelligence Free; Agency Is the New Wealth Differentiator — 4KTV · 2026-08-09
- Anti-Hustle Culture: Long-Term Dominance Relies on Non-Linear Sustainability — fkasummer · 2026-08-09
- Jeff Ladish: Without International Coordination, Consequentialist AI Will Become Schemers — JeffLadish · 2026-08-09