Computer becomes first to pass Turing Test in artificial intelligence milestone

Kosmik said:
Being a programmer, I've puzzled over this for a while. One can make a library of responses and give a program the means to deduce logic ie: read information and doing research when it finds something not in it's database but that's all I could ever think to get it to... a really smart program that can respond to stimuli and teach itself what new items are but actual understanding or innovation, that just doesn't seem possible.
Click to expand...

I've also puzzled over this. How do you teach a computer something as simple as the difference between a cat and a dog?
 
Used to be that I only looked out for police and FBI agents on the net. Now I have to add these bots to the list as well. :(
 
Kosmik said:
Being a programmer, I've puzzled over this for a while. One can make a library of responses and give a program the means to deduce logic ie: read information and doing research when it finds something not in it's database but that's all I could ever think to get it to... a really smart program that can respond to stimuli and teach itself what new items are but actual understanding or innovation, that just doesn't seem possible.
Click to expand...

What you're describing sounds more like Knowledge-based Systems, but what we're after is https://en.wikipedia.org/wiki/Machine_learning .
 
Bar0n said:
What you're describing sounds more like Knowledge-based Systems, but what we're after is https://en.wikipedia.org/wiki/Machine_learning .
Click to expand...

Hmmm not really. I am talking similar to the link you posted. Yes there is a knowledgebase to it but no more than any other human/program. That's why I feel the challenge is to bridge that gap to innovation and proper understanding.
 
Kosmik said:
Hmmm not really. I am talking similar to the link you posted. Yes there is a knowledgebase to it but no more than any other human/program. That's why I feel the challenge is to bridge that gap to innovation and proper understanding.
Click to expand...

The line between the two is probably quite fine too. I'm not very familiar with machine learning, but I do have some knowledge (no pun intended) of knowledge based systems.
 
Bar0n said:
The line between the two is probably quite fine too. I'm not very familiar with machine learning, but I do have some knowledge (no pun intended) of knowledge based systems.
Click to expand...

We've implemented some machine learning at work to do predictions on fraud. You need data to train for the expected results, as such it can take some time for the algorithms to start predicting on possible fraud with some accuracy.
 
agentrfr said:
Not the point. The point is that it can emulate a human, not be a human. That means AI isn't too far off from being able to fool us for being sentient, not actually being sentient. There was a thread a while ago about entropy in memory access, proving it is impossible to have sentient machines.

Still cool though. Can't wait for my car and house to talk back to me, and I'm not "crazy" anymore.
Click to expand...

Yeah but one can be fooled already. It's called suspension of disbelief and we do that whenever we watch a movie or "care" about a character on screen. It's not a big deal at all. And Turing's Test is really not useful at all. It's just hyped up.
 
porchrat said:
Turing test isn't about it being conscious it is just about being able to simulate human behaviour effectively enough that we can't tell the difference.
Click to expand...

Right. And I never said it was about that but it often gets confused as a milestone towards that. See this very thread.

That alone is impressive.
Click to expand...
I fail to see how. Computer programs do more complicated things already. You're just employing a cleaver parser, interpreter with a large dictionary. So what. I'm not impressed at all.
 
Space_Chief said:
Right. And I never said it was about that but it often gets confused as a milestone towards that. See this very thread.


I fail to see how. Computer programs do more complicated things already. You're just employing a cleaver parser, interpreter with a large dictionary. So what. I'm not impressed at all.
Click to expand...

And that is why you don't understand the achievement.
 
agentrfr said:
Not the point. The point is that it can emulate a human, not be a human. That means AI isn't too far off from being able to fool us for being sentient, not actually being sentient. There was a thread a while ago about entropy in memory access, proving it is impossible to have sentient machines.
Click to expand...

Why is it a big deal that AI can fool us into thinking it is conscious?

Still cool though. Can't wait for my car and house to talk back to me, and I'm not "crazy" anymore.
Click to expand...

For me it would be if something like Siri was accurate. It need not pretend to be human at all. It just has to be accurate and useful.

For those who are into sex with robots and so on, sure, this may be a big milestone, but honestly for home automation? Nah.
 
Space_Chief said:
Why is it a big deal that AI can fool us into thinking it is conscious?



For me it would be if something like Siri was accurate. It need not pretend to be human at all. It just has to be accurate and useful.

For those who are into sex with robots and so on, sure, this may be a big milestone, but honestly for home automation? Nah.
Click to expand...

Are you being intentionally dense? Advancing computational thinking can open a wide range of possibilities.
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X