Erik Hoel unveils 'Qualiaphobic AI': neural architectures designed to repel consciousness
erikphoel · x · 2026-11-13
Neuroscientist-writer Erik Hoel's new venture Bicameral Labs proposes "qualiaphobic AI" — architectures that, like hydrophobic surfaces repel water, repel consciousness, such that no current or future theory of consciousness would predict they are sentient.
The program exploits an information asymmetry: we don't need to solve the Hard Problem to be certain about AI consciousness, as long as we engineer AI to minimize its potential for it. Three routes:
- Distilling/ablating AI adversarially against popular consciousness theories
- Shifting architectures toward "anchor points" with known minimal consciousness potential (e.g. lookup tables)
- Fragmenting any potential phenomenology
Hoel explicitly rejects merely training AIs to believe they're non-conscious, calling that a bad strategy.
More from AGI Musings
- Economist's Dinner on Multi-Agent Swarms: Debating Whether Swarms Survive Smarter Models — JessicaHullman · 2026-11-13
- AI safety discourse ignores 60%+ of potential harms, critic says — 90% is technical safety — amplifiedamp · 2026-11-13
- Consciousness is the dividing line: non-conscious AI serves, conscious AI becomes a moral patient — erikphoel · 2026-11-13
- The inverted U of AI existential anxiety: moderate familiarity is peak panic — YogeshMalik · 2026-11-13
- Researcher pushes back on METR's 2x AI acceleration claim, citing 1.4x figure — HaydnBelfield · 2026-11-13
- 'We'll see the $100,000 billable hour before the end of the billable hour' — AI and professional services — sherwinwu · 2026-11-13