rgblong: real risk lies in instilling contradictory views of consciousness and goals in models
rgblong · x · 2026-09-18
Continuing the debate, rgblong clarifies his specific safety concern: embedding contradictory or incoherent views about consciousness, goals, and self into models. It's an addendum to the discussion of Microsoft's stance on model self-presentation.
Related event: Microsoft's 'Model Welfare' Push Sparks AI Safety Debate(6 posts)→
More from AGI Musings
- Boss AI makes agent teams pad reports and cost 1.5x more, study finds — i_dg23 · 2026-09-18
- AI researcher: generative models will do to math what recording did to music — aminkarbasi · 2026-09-18
- Stop glorifying IMO medalists: AI talent worship deserves scrutiny — tak3sh8 · 2026-09-18
- DeepMind's Stephan Hoyer: high-fidelity physical world simulation is nearly impossible — DaniloJRezende · 2026-09-18
- Plinz: No one will pause AI — the real goal is keeping capable models away from the public — nptacek · 2026-09-18
- Debate: Should future people be discounted? Longtermism vs temporal discounting on X — NathanpmYoung · 2026-09-18