You know your own experience directly. But everyone else? You only ever observe their behavior — never their inner experience. The other minds problem asks how you can justify believing other people are conscious at all. With AI, this ancient puzzle gets teeth.
The leap of faith you already make
I have a mind. Others act like me. So... others probably have minds too.It's an inference from analogy, not proof. You assume other people aren't empty philosophical zombies — and almost certainly you're right, because they share your biology and evolution.
Why AI breaks the usual shortcut
The analogy that works for other humans fails for AI: a model can act conscious — claim feelings, plead not to be shut off — while sharing none of our biology. So:
- We can't use behavior alone (The Turing Test, The Chinese Room).
- We have no consciousness-meter (The Hard Problem of Consciousness; IIT is a contested attempt).
We may build AIs that pass every test for a mind and still never know if anyone's home — which makes Moral Patienthood and mind crime decisions we'll have to make in the dark.
Related: The Hard Problem of Consciousness · Mary's Room and Philosophical Zombies · Moral Patienthood