Friday Checks: of 11 viral AI claims, only 5 turned out true
eyishazyer · x · 2026-11-07
AI researcher eyishazyer is launching a recurring "Friday Checks" series, verifying viral AI claims against primary sources each Friday. This week's batch of 11: 5 true, 5 misleading, 1 false.
Verified examples:
- True — "Astra edited a model's weights, math up 6x": real, but only demonstrated on a tiny 0.8B model
- Misleading — "Democrats popped the AI bubble": chip earnings lit the fuse first
- Misleading — "Grok bot quietly cut its limits": Musk blamed costs, but no cut is confirmed
- True — "AI found 948 Linux kernel bugs": confirmed per OpenAI
Related event: Friday Checks: Only 5 of 11 Viral AI Claims Hold Up(3 posts)→
More from AGI Musings
- Anti-AI data center backlash mirrors the costly anti-nuclear hysteria of the 70s — QuintinPope5 · 2026-11-07
- Nous Research releases Arete-27B, cutting a Chinese open model's propaganda rate from 22% to 1% — NousResearch · 2026-11-07
- AI-generated TikTok stories are surprisingly reactionary, observer notes — aronchick · 2026-11-07
- Essay "AI Needs Soft Power": the industry builds tech but neglects public understanding — RichardsonDx · 2026-11-07
- Uncle Bob says deep language expertise is now a waste, but can junior devs copy that? — bendee983 · 2026-11-07
- When AI grants every wish, genuine uncertainty may become the ultimate luxury — VraserX · 2026-11-07