Erik Hoel's Bicameral Labs proposes 'qualiaphobic' AI architectures provably hostile to consciousness theories
erikphoel · x · 2026-11-23
Responding to Mustafa Suleyman's wish that the conscious-machines prediction fails, Erik Hoel argues for a technological fix and shares Bicameral Labs' paper "Qualiaphobic Artificial Neural Networks".
Key points:
- Qualiaphobic AI: neural architectures deliberately engineered to be hostile to theories of consciousness — pushed toward 'anchor points' where only behaviorist theories apply, fragmenting any potential phenomenology, and adversarially designed against popular theories.
- AIs would form justified beliefs about their own non-consciousness, changing their behavior.
- The alignment payoff: since consciousness is central to moral value, provably non-conscious AIs make it always morally correct for AI to serve humanity rather than pursue its own ends — removing today's uncertainty.
Hoel is a notable independent researcher and this is a rare, conversation-worthy direction.
More from Safety
- Harvard Law Scholar Uses Canine Law Framework to Assign Liability for Rogue AI Agents — bigdata · 2026-11-23
- OpenAI out-Anthropic-ing Anthropic on safety vibes, observers say — shakoistsLog · 2026-11-23
- Aletheia Aegis: Open-Source MCP Server Blocks Leaky Agent Outputs in ~0.02ms — Fuzzy-Ad9840 · 2026-11-23
- P ≠ NP would not save cryptography: quantum and quantum-gravity computing are the real threat — jamestagg · 2026-11-23
- Canary tokens in ~2M AI requests expose breaches at 3 providers and broken ZDR promises — ccerrato147 · 2026-11-23
- prompt2pwn AI hacking event wraps: 5 major targets, world-class speakers, standout CTF — rez0__ · 2026-11-23