• Welcome to the new Internet Infidels Discussion Board, formerly Talk Freethought.

Historic Achievement: Microsoft researchers reach human parity in conversational speech recognition

I did not say it need to understand, I said it has to look like it understands.

We don't want robots to pass the Turing test. That would be worthless to us. We just want them to make us happy and fix shit for us.

I think you have some sort of fantasy/sci-fi world concept of AI.

But one way of looking like it understands is actually understanding.

No, it's not. It's more a measure of how easily fooled you are. Understanding isn't a question of appearance. It's like saying that the important thing when passing a test is to look competent, not that you know wtf you're doing.

But your google argument is irrelevant because as long as machine passes the turing test for all intents and purposes it has a mind, certainly in practical sense.

Well, I agree with you. But it's highly controversial, and hinges on how we define mind. I have a very wide definition of mind. I'm on team Dennett. He defines mind as a collection of sensory inputs and outputs. It's the "nervous system" that makes something a mind. Not necessarily the thinking.

BTW, "passing the Turing test" is something different than something being "Turing complete". The Turing test is pointless, while Turing complete is a great goal, but something we will never reach.
 
I really don't see a point here. It seems you are arguing for the sake of arguing.
As I said if AI is to replace human then it has to act like human and and be good at it. Right now it's not. Regarding speech recognition it's still far from humans.
 
Other than the fact that AI isn't meant to act like a human and it doesn't replace humans, it augments human endeavor, this discussion has been interesting. The issue DrZoidberg and I were beginning is whether AI is superior as an instrument for accomplishing material tasks to man. I say it is because it is , can be, demonstrably superior at accomplishing tasks. DrZoidberg suggests AI has been captured by those who choose to have computers replace humans in some, mostly dangerous, cognitive tasks. Obviously I think differently.
 
Last edited:
I really don't see a point here. It seems you are arguing for the sake of arguing.
As I said if AI is to replace human then it has to act like human and and be good at it. Right now it's not. Regarding speech recognition it's still far from humans.

I suspect it's more a question of you not understanding the words you are using (in this context).
 
BTW, "passing the Turing test" is something different than something being "Turing complete". The Turing test is pointless, while Turing complete is a great goal, but something we will never reach.

I think you're misunderstanding Turing completeness - it's not a quality of artificial or natural intelligences, or even necessarily desirable. Turing incomplete systems can still be useful to humans, and some Turing complete systems not so much (beyond entertainment or curiosity).
 
I really don't see a point here. It seems you are arguing for the sake of arguing.
As I said if AI is to replace human then it has to act like human and and be good at it. Right now it's not. Regarding speech recognition it's still far from humans.

I suspect it's more a question of you not understanding the words you are using (in this context).
It's not me, it's you.
 
Back
Top Bottom