Reply to Bengio: we never solved 'human alignment' before demanding AI alignment
kwangmoo_yi · x · 2026-09-12
Replying to Yoshua Bengio, researcher kwangmooyi argues we may be thinking about the wrong problem: before 'AI alignment,' we should recognize humanity never succeeded at 'human alignment.' If these models are (or become) intelligent entities, he suggests the alignment problem may not be easier than our own.
Related event: Researcher Pushes Back on Bengio: Solve Human Alignment First(2 posts)→
More from AGI Musings
- 24 Fields Medalists Issue Declaration Warning of Severe AI Misalignment in Mathematics — kimmonismus · 2026-09-12
- 24 Fields Medalists Speak Out on Severe AI Misalignment in Mathematics — kimmonismus · 2026-09-12
- What Good Is an AI Math Proof Nobody Understands? The Understanding-Lag Debate — ctjlewis · 2026-09-12
- Counter to 'AI won't gain power' argument: AIs can replicate thousands of times, never sleep — AndyMasley · 2026-09-12
- What if the most important work was never a job? Rethinking human value after AGI — jeremy_goldberg · 2026-09-12
- Richard Socher: advance science safely with AI, ignore sci-fi doom scenarios — RichardSocher · 2026-09-12