OT - Sentience

[narf poit chez BOOM]

I have studied learning AI; every single possible path the AI may learn and grow in is implicit in the algorithms the AI uses to learn and the data set provided to learn from. If you iterated over X generations exploring every possibility, you'd have every possible X generation AI.

A self-growing gear box is still a gear-box.

(This is why I prefer the term VI - Virtual Intellgence)

[Ed Kolis]

I agree that we're not there yet in terms of the soldiers in Call of Duty or the dwarves in Dwarf Fortress, but I'm saying that it's within reason to think that we could be there in the next, say, 25 years - sooner if you include "the Internet as a whole" among things that you might consider self-aware.

[Raapys]

I don't think the current computer system has what we need, since it's limited by its programming. While it might be possible to make an AI that writes new and changes old code all by itself, even that might not be enough to get us what we want.

I think the best solution would be to instead design a new computer system which would try to directly replicate the way the brain does things.

Interesting subject though, in these days where direct brain-to-computer interaction is actually becoming a reality( Link ) and where it has been discovered the brain has already made your decisions long before your conciousness becomes aware of it( Link ).

[narf poit chez BOOM]

The video in the first link is vague and indefinite, with poor sound quality.

The second link is interesting, although lacking detail.

[Ed Kolis]

Honest, officer, I didn't plan to kill him... my brain made me do it!