Will AI Become Conscious?

Frame the driving question and surface the contradiction: AI sounds more human than ever, but 'sounding like' something and 'being' something have always been different things.

• The driving question, stated plainly
• Why this is harder than 'will robots take our jobs'
• Two failure modes: false confidence in either direction

Lock in the learner's intuition before any evidence is shown, so the investigation has something concrete to push against.

• One independent choice, no hedging
• The guess becomes the reference point for the rest

Concrete evidence: token prediction, training on human text, no persistent memory between sessions, and no internal world model in the human sense.

• Autocomplete at enormous scale
• No persistent self across conversations
• Behavior driven by statistical patterns, not goals or feelings

A hands-on comparison widget: rewrite prompts and see how a model adjusts tone, emotion, and 'opinions' — exposing that the apparent personality is a controllable output, not a fixed inner state.

• Tone changes instantly with prompt framing
• The 'self' shifts to match the user
• Convincing voice ≠ stable identity

Lay out the three serious scientific accounts of consciousness — Global Workspace, Integrated Information Theory, and higher-order thought — and what each would require from a machine.

• Global Workspace: broadcast and integration across modules
• IIT: phi, the amount of integrated information
• Higher-order thought: a system that models its own states
• No current AI architecture clearly satisfies any of them

Name what we genuinely do not know: substrate independence, whether consciousness requires biology, and the hard problem itself.

• The hard problem is unsolved for humans too
• Substrate independence is a bet, not a fact
• A lack of consensus is itself a finding

Transfer test: the learner evaluates a hypothetical future system against the three theory-derived criteria and decides whether the claim of consciousness is warranted, justified, or premature.

• Use the criteria, not feelings
• Justify the verdict
• Recognize what evidence would change the answer

Directly answer the driving question: today's AI is not conscious, future AI may or may not be, and the only honest position is to require evidence against criteria — not vibes.

• Not conscious today, by every serious test
• Future is open, not predetermined
• The right move is criteria-based evaluation