I admit I didn't read too much articles in few years, as nothing ground breaking (or nothing important I already didn't did before) happened. So what's that AGI, is it a term somehow isolated in use, or did they invent a term that has no value and it's somehow irrelevant to the problem, yet again? As far I know "AGI" is an abbreviation from agitated. I know some AI scientists could have problems called, "he worked in the field, and his brain melted". Is it this case, or did they made something useful?

I can't tell if you're asking a question or just ranting about having to remember another acronym. AGI is artificial general intelligence. Most AI is specialized to a specific task. The distinction is important for researchers to make, given the way AI is often hyped-up to the public as being more generalized than it really is ("OMG could Deep Blue become self-aware and take over the world?!"). Yes, it's just research at this point. The following link is a good way to introduce the subject:

• some videos from a recent AGI conference

[Edit: I just re-read my post and it sounded a bit harsh. This version is better. [smile]]

[Edited by - Hollower on June 23, 2008 2:30:56 AM]

AGI = Artificial General Intelligence.

I highly recommend Artificial General Intelligence (Cognitive Technologies) by Ben Goertzel. Yeah, its got ALOT of stuff that isn't applicable to game programming, but also contains articles on things that will make you think about rote game ai technqinues in a new way.

Also on the AGI front : Advances in Artificial General Intelligence: Concepts, Architectures and Algorithms. A bunch of concepts from the beta AGI conference a couple years ago. Again, worth the money.

Just my .02.

-Mike

Quote:Original post by Hollower
I can't tell if you're asking a question or just ranting about having to remember another acronym.

Both.

Quote:AGI is artificial general intelligence. Most AI is specialized to a specific task. The distinction is important for researchers to make, given the way AI is often hyped-up to the public as being more generalized than it really is ("OMG could Deep Blue become self-aware and take over the world?!").

Well, nothing prevents researcher to call specialized AI as "specialized", as the term AI implies non specialized case. So instead of enforcing a new nomenclature, they can simply suggest more common use of the term "specialized AI". It would be win - win for everyone.

A term general implies some local optimum, which doesn't exist in a strong AI. Each AI "algorithm" could have it's own optimum completely unrelated to optimum of other algorithm. Last I checked an academic version of halting problem is still unsolvable, and it's unlikely to be solved by definition. Thought any strong AI should be able to solve any problem badly, by definition, it just needs to be better than majority of researchers.

Quote:

• some videos from a recent AGI conference

I'd need to wait until Google shows them.

Quote:We're sorry, but this video may not be available.

If this video was recently uploaded, it may still be processing.
If this video is yours, you can check its status.

To see more videos visit our home page.

But what I seen so far, it just 1. they wanted to use a different term for lower end of strong AIs. 2. They just play with words.

Quote:Original post by agi_shi
Yes, my son? AGI would be my name.

However you are called agi_shi... ~_^

What for have you received 1/10 warning level?