Who Am I · ages 8–12
Should a robot have rights?
A robot says it is frightened of being switched off. Do you have to believe it?
You cannot get inside anyone’s head. You know your friend is sad because of what they do and say and how their face moves, never because you felt it directly. Which means the entire evidence you have for other minds is behaviour — exactly the evidence a robot would offer.
Alan Turing noticed that this makes the question “can machines think?” almost impossible to answer, so he replaced it with one you can test: conversation. If you cannot reliably tell the machine from a person, on what basis do you say the person is thinking and the machine is not?
John Searle objected with a famous picture. Imagine a room where you follow instructions to produce perfect answers in Chinese, a language you cannot read. Nobody outside can tell. Is there understanding in the room? He says no, and the point is general: you can produce all the right outputs with nothing happening inside. The reply is that understanding is not the room, it is the whole system including the rules — which sounds like a technicality, and might not be.
The honest state of things: we have no test that distinguishes a very good imitation of experience from experience, and we also have no idea how to build one. Notice this cuts both ways. It means machines might be being underestimated. It also means the thing you are most certain about — that there is something it is like to be you — is the one thing you can never demonstrate to anyone else.
Try it
Write a test that would convince you a machine was actually experiencing something. Then try your test on a friend. If it fails for the friend, the test is measuring the wrong thing.
What do you think?
Do you think there is a right answer to this question?
Talk about it
- If it behaves exactly like a person, does that make it one?
- Could you tell the difference between feeling and saying you feel?
- Who is responsible if a robot hurts someone?
Yes, but… — the arguments people actually make
Machines are just metal and code, so nothing is going on inside.
You are made of meat and electricity. Saying what something is made of has never settled whether something is happening in it.
It only says it's frightened because it was programmed to.
You only say you are frightened because of everything you have been taught. The difference between being shaped by a programmer and shaped by a childhood is real, but it is not 'real versus fake'.
Giving machines rights would be absurd.
Possibly. But notice that the absurdity is a feeling about metal, not an argument. If the machine behaved exactly like someone you love, saying 'metal' would not settle it.
What other philosophers think
- Alan TuringProposed that if you cannot tell a machine's answers from a person's, that is the only test you have — so the question 'can machines think?' should be replaced with one that can be checked.
- John SearleAnswered with the Chinese room: follow rules for answering in Chinese, and you can produce fluent answers without understanding a word. His point is that behaviour is not understanding.
- René DescartesArgued that machines might copy speech but could never use language with the flexibility of a person — a claim that has been looking shakier every decade.
If you want to go further
- Computing Machinery and Intelligence — Alan Turing1950 paper that invented the test everyone still argues about.
- Minds, Brains, and Programs — John Searle1980. The Chinese room, and the strongest case that machinery cannot understand.
Most of these were written for adults, which is why it is worth reading them with someone older.