[edepot]

I think the main problem with AI is that they are not self-aware.  To achieve self-awareness, you need a virtual machine and feedback loop.  This allows for thinking machines and new behavior.  This is a complicated subject, but for those who are interested, the reasons for these prerequisites are defined discussed in this article:

http://www.edepot.com/forums/viewforum.php?f=12

Finally we are able to achieve self-awareness without resorting to state-machines and predefined scripts.

[Maviarab]

Welcome to the site edepot.

No need to think, the fact that current Ai is not self aware 'is' the problem. Bu then as we have discussed many times, and think about this very carefully, would you relly want a machine that is truly 'self aware' ?

[Freddy]

Welcome to the forum from me too.  You look like a good writer - if you ever want to publish an article here or work on the AI Wiki let us know.

[Art]

Welcome to the playground!

Mav, I read in his 6th paragraph that the "behavioral" side would not be altered but rather
the side that allowed for learning and thinking would be affected.

IMO, that would certainly be a much better approach and more acceptable.

Nice work!

[Dante]

This 'is' kinda being worked on.


Look up; AGI (artificial general intelligence)

and mostly related; semantic web.

[lrh9]

I've heard the term feedback loop used a lot. What do you mean? Is it a part of the a.i. program where the output data from previous operations become input data?

[Freddy]

Well in electronics a feedback loop IS where some of the output gets sent back to the input...so I think you may be right that this is something similar.

[kaat]

I assume you're referring to this post: http://www.edepot.com/forums/viewtopic.php?f=12&t=1534

Interesting article! Just musing aloud ..

It seems there's a fundamental question here about our expectations of an intelligent system. To what extent are WE able to modify our own instruction set? If we're talking about everyday thoughts and behaviors, obviously we do well. But if we talk about our genetic tendencies, our core psychology, the lower brain, etc. - it's a lot more hit and miss. We can't even be sure if we have a prime directive, or what it is. I seriously question your assertion that it's simply "survival". Obviously it gets very philosophical at this point, but I would say my desire for truth and meaning trumps my survival instinct. Survival, we know about; truth and meaning - not so much.

Can we create a system that overcomes these issues? I would say yes. But then, what is the prime directive of this system? Or is there one? Is the intent that this system will go beyond our intelligence? Isn't this a lot like our desire to be more God-like? What truths might such a system reveal? 42?

It seems to me that if we create a system that is equal to the human brain, it will have an equal chance of going terribly wrong. And we'll be furiously feeding it all our wisdom to help it avoid going wrong, and it'll still go wrong. Trouble is, how are we defining wrong at this point? Of right for that matter?

[kaat]

I have the feeling that if this is about our own progress, we might be better off looking into artificial dumbness. I'm serious.

[lrh9]

I just want my a.i. to obey my commands.

[kaat]

I just want my a.i. to obey my commands.

Hi lrh9,

I'm intrigued by your comment. But if you're not into pulling it apart, of course I respect that.

[lrh9]

Not everything about a.i. is complicated. If I tell my a.i. to bake a cake. It should attempt to bake a cake. A.I. should do what people tell a.i. to do.

[kaat]

Not everything about a.i. is complicated. If I tell my a.i. to bake a cake. It should attempt to bake a cake. A.I. should do what people tell a.i. to do.

But if you were asked to provide AI for a 'general purpose' system, what would you tell it to do? Perhaps simply 'learn my commands'? What's the first command?

[lrh9]

What?

[Dante]

Now, if we could answer that....