The Ziff argument
"What we find in our laboratory is this: no robot could sensibly be said to feel anything.
Why not? 9. Because there are not psychological truths about robots but only about the human makers of robots. Because the way a robot acts (in a specified context) depends primarily on how we programed it to act. Because we can program a robot to act in any way we want it to act. Because a robot could be programed to act like a tired man when it lifted a feather and not when it lifted a ton. Because a robot couldn't mean what it said any more than a phonograph record could mean what it said. Because we could make a robot say anything we want it to say. . . .
Because no robot would act tired. Because a robot could only act like a robot programed to act like a tired man. For suppose some robots are programed to act like a tired man after lifting a feather while some are so programed that they never at like a tired man. Shall we say "It is a queer thing but some robots feel tired almost at once while others never feel tired"? Or suppose some are programed to act like a tired man after lifting something blue but not something green. Shall we say "Some robots feel tired when they lift blue things but not when they lift green things"? hard work makes a man feel tired: what will make a robot act like a tired man? Perhaps hard work, or light work, or no work, or anything at all. For it will depend on the whims of the man who makes it"
(P. Ziff, 1959, p. 103).
References
Ziff, Paul. 1959. The feelings of robots. Analysis, Vol. XIX, No. 3. Reprinted in Minds and Machines (1964). Alan Ross (Ed.), pp. 98-103.
|