Erik Hoel proposes 'qualiaphobic AI': neural nets engineered to minimize machine consciousness
erikphoel · x · 2026-11-13
Author-researcher Erik Hoel published an arXiv paper introducing 'qualiaphobic AI': neural network architectures deliberately hostile to theories of consciousness, aimed at minimizing the potential for artificial consciousness. The paper justifies qualiaphobia via the weakness of current consciousness theories, lays out three converging engineering approaches (approaching 'anchor points' of minimal-consciousness systems, fragmenting potential phenomenology, adversarial design), proposes concrete qualiaphobic transformations toward behaviorism-only systems, and explores how AIs could form justified beliefs about their own non-consciousness and how that would change their behavior.
Related event: Erik Hoel Proposes "Qualiaphobic" AI Safety Architecture(8 posts)→
More from AGI Musings
- tszzl: local minima are rare in high-dimensional spaces — the greatest whitepill — JensHonack · 2026-11-13
- Ted Chiang Quote Recirculates: Fear of Tech Is Fear of Capitalism Using It — round · 2026-11-13
- Matteo Pasquinelli's four theses on AI's place in the history of technoscience — ArtificialOther · 2026-11-13
- Behzod Sirjani launches Yet Another Magazine to cover AI's missing human layer of work — beh_zod · 2026-11-13
- Replit CEO Amjad Masad: AI writing more reliable code than humans is dangerously close — VraserX · 2026-11-13
- Essay: the technical moat is drying up — what software engineers should do now — dotey · 2026-11-13