Researcher: alignment will be solved the bitter-lesson way, not via human enhancement
tszzl · x · 2026-11-03
AI researcher tszzl argues that solving alignment has always been about following the same bitter-lesson-driven scaling path as the rest of modern AI — not waiting around for "human enhancement" or similar approaches.
More from AGI Musings
- Students keep a secret doc of real work because a teammate keeps replacing it with AI slop — IanArawjo · 2026-11-03
- If alignment faking exists, could reverse stress testing reveal misalignment faking? — eigenron · 2026-11-03
- Audit picks the sycophant model 3/3 times, but "telling apart isn't explaining" — kevrussell · 2026-11-03
- The real Minecraft agent benchmark: can a kid have fun playing with it live? — akbirthko · 2026-11-03
- Andrew Ng's Batch Halloween issue: AI witch hunts and out-of-control agent fears — DeepLearningAI · 2026-11-03
- Bay Area machine shops field 5-10 calls a week from AI manufacturing startups — NYCounihan · 2026-11-03