AI – The paradox

Good point expliciting the "paradox". But I don't think we need to change the definition of AI, or to add a definite purpose. Currently, "successful" AI systems are indded systems with a purpose: cheating some human judges into beliveing your a boy in a chat, wining a Jeopardy game, detecting cats in video. Well done. But lame. With a purpose you get working systems, but you don't get what we expect when we say we want an AI.

If you defined the purpose to be to execute any order given. by a human, it would be very nice, but it still wouldn't be enough.

Perhaps the purpose would be to find out why we built them? In a way, it's a paradoxical question, like whether God exists, or why we are here? I guess it would do, as a purpose, to obtain the wanted AI. I'm not sure it would help the engineers to build it, having such a purpose.

The purpose is whatever the AI is designed to do. E.g. pass a Turing test, design things, whatever.