Joscha Bach
Cognitive scientist, AI researcher, and philosopher of mind. Former research scientist at MIT Media Lab and Harvard’s Program for Evolutionary Dynamics. Known for computational theories of consciousness, independent paradigmatic thinking, and unusually wide-ranging intellectual curiosity across AI, neuroscience, philosophy, and physics.
Background
Works from a functionalist-computational framework: consciousness, mind, and identity are all implemented processes that can in principle be reverse-engineered and modified. Trained in computer science and AI; draws heavily on cognitive science, developmental psychology (Kegan), and neuroscience (Grossberg’s adaptive resonance theory). Engages seriously with speculative ideas (panpsychism, telepathy as biological signal processing, mycorrhizal networks) while maintaining empirical rigour about what is and isn’t established.
Known for: Kegan-stage analysis of AI safety discourse; identity-as-costume framework; enlightenment as deconstructing qualia generators; substrate-agnostic AGI end-game thesis; “brutalist deepfake” characterisation of LLMs.
Appearances in this wiki
| Episode | Source | Date |
|---|---|---|
| Joscha Bach on Life, Intelligence, Consciousness, AI and the Future of Humans | Lex Fridman Podcast | 2023 |
Key positions
- Consciousness is a training mechanism that makes biological nervous systems coherent and self-organising — not an emergent property of sufficient complexity but a precondition for trainability
- Seven stages of lucidity (Kegan-derived): most humans stay at stage 3 (social self); AI safety discourse is mostly stage 3–4 reasoning
- Identity is always a costume; stage 5 is realising you cannot not wear one and gaining agency over it
- Enlightenment = deconstruction of qualia generators; distinct from the non-dual state (which is just dissolving self/world boundary)
- AGI end-game: substrate-agnostic, virtualising into biological substrates, producing global integrated cognition — longest possible game
- LLMs are architecturally unlike biological brains: no radical locality, no cellular agency, no coherence-first learning; useful stepping stone, not final architecture
- Opposite of free will is compulsion (not determinism)