How to Build Conscious Machines

Proposed hierarchy & dissolving the hard problem

• Discussion centers on a five-stage hierarchy of consciousness, from inert (0) through hard-coded, learning, first-order self, second-order selves (access consciousness, theory of mind), to third-order selves (modelling others’ inner dialogues).
• Some see this as a promising, evolution-grounded reframing that treats phenomenal consciousness as functional and rejects philosophical zombies.
• Others argue it “defines the hard problem away” by assuming qualia are reducible / functional without really explaining why subjective experience exists at all.

Inner narrative, aphantasia, and varieties of cognition

• Long subthread on people without inner monologue and/or mental imagery (aphantasia, anendophasia).
• Reports of rich reasoning and social prediction (“just knowing the answer”) without verbalized thoughts challenge tying higher stages strictly to linguistic inner narrative.
• Counter-claims: inner speech is ubiquitous but under-noticed; cognition is mostly pre-linguistic even for people with strong monologues.
• Raises classification questions: how to place non-verbal but clearly capable humans in the hierarchy?

Animals, play, and theory of mind

• Debate over whether cats/dogs/crows fit second- vs third-order levels.
• Examples of animal play (e.g., “snowboarding” crow) and anticipatory behavior (dogs preventing accidents) are cited as evidence of projection and perhaps joy.
• Many see consciousness as a spectrum with no hard boundaries; warnings against overfitting human traits (especially language) into the definition.

IIT, substrate independence, and AI

• Interest in Integrated Information Theory as a rare mathematically explicit theory; strong criticisms about misunderstandings of computation and Turing universality.
• Disagreement over whether substrate matters: some argue brains and LLMs are fundamentally different; others emphasize functional/behavioral equivalence and substrate independence.
• Questions raised: if phenomenal consciousness is inherently functional, must advanced LLMs count as conscious? No consensus.

Qualia, physicalism, and panpsychism/idealism

• Dispute over whether qualia are irreducible “atoms” of experience vs abstractions built from more basic processes.
• Some favor physicalist, emergent accounts; others lean toward panpsychism or idealism and doubt any information-processing theory can capture “what-it’s-like-ness.”

Usefulness and ethics of conscious machines

• Practical camp asks: is consciousness even useful for AI (e.g., lawyering), or just an accidental byproduct?
• Concerns that creating genuinely conscious machines mainly risks suffering/slavery; suggestion that “we can already build conscious systems called babies.”
• Meta-critique: thesis feels meme-y and jargon-heavy to some, though underlying work is peer-reviewed.