Reddit Reflection: Altman's 'Save RAM for AI' Comment and the Alignment Paradox
mboazak · reddit · 2026-11-11
A Redditor reflects on Sam Altman's suggestion that humanity should prioritize resources for AI: in an energy-scarce society, putting the tool above the people it serves is backwards. The deeper worry is misalignment — given infinite resources to solve all humanity's problems, one time in ten the tool may reasonably conclude that humanity itself is part of the problem. Full-steam RSI defies alignment by nature, and the very attempt to align agents may backfire. The author self-deprecatingly admits to still refreshing OpenRouter waiting for 6.2 — pure copium.
More from AGI Musings
- Frontier open and closed models are only 6-12 months apart, says Stanford's Mirhoseini — agihouse_org · 2026-11-11
- Azalia Mirhoseini: model layer will commoditize, custom chips face a Cambrian explosion — agihouse_org · 2026-11-11
- AI Beating Nearly All Scientists in 2-3 Years? Hard Sciences Have Physical Limits — niloofar_mire · 2026-11-11
- OpenAI researcher says AI will best all but 97th-percentile scientists in 2-3 years — willdepue · 2026-11-11
- Researcher pushes back on OpenAI's Depue: many top AI researchers come from the third world — niloofar_mire · 2026-11-11
- OpenAI researcher predicts AI will beat 97th-percentile scientists within 2-3 years — willdepue · 2026-11-11