Author defends quialiaphobia: designing AI to avoid consciousness may scale with capability safeguards
erikphoel · x · 2026-11-13
The author responds to the main objection to his paper on quialiaphobia — deliberately designing AI architectures to avoid consciousness — namely that it may cost too many resources or hurt capabilities:
- It might turn out easy; we won't know how close we can get until we try.
- The three approaches (shifting networks toward anchor points with minimal potential for consciousness, phenomenological fragmentation, adversarial design against existing theories of consciousness) may exhibit strong consilience and scale together. If the system is close to a lookup table in key respects plus assurances from the other methods, an AI could form justified beliefs in its own non-consciousness.
- There may be forceful moral reasons to accept the capability limitations this requires — and slowing capabilities isn't bad for an AI safety plan.
The excerpted chapter notes a qualia-phobic AI would fit no current theory of consciousness and possess no human-conceivable phenomenology; combined with alignment techniques already used to instill model personality, this could yield most benefits of a provably non-conscious AI world.
More from AGI Musings
- Analyst: Everyone Underestimates Non-AI R&D Acceleration — It'll All Be Ripping Anyway — teortaxesTex · 2026-11-13
- Sentience Institute validates a psychometric scale for 'substratism' — moral bias against AI — jacyanthis · 2026-11-13
- Anthropic's alignment work paying off as OpenAI falls behind, argues researcher — haider1 · 2026-11-13
- Claim: Senior researchers at OpenAI and Anthropic have confirmed contested study result — NathanpmYoung · 2026-11-13
- Investor tszzl: Chinese open-source models have earned huge global goodwill and soft power — tszzl · 2026-11-13
- 'We Have a Right to Compute': Critics Warn US AI Regulation Follows Nuclear Energy's Playbook — nptacek · 2026-11-13