Wednesday, December 2, 2009

Advantage of the unborn. AI and self-replication

A.I. (artificial intelligence) is not yet where we want it to be (at least where those of us who grew up reading Asimov would like it to be). That's for sure.
It's also not yet where many expected it would be by this time. We tend to underestimate the complexity of the problem of reaching a level of intelligence comparable to ours. -although we can lower the bar a lot with our own stupidity from time to time-
But although it has a long way to go, we could say it is slowly moving forward through different fields. Like a child who is slowly learning in an extraterrestrial world.
We humans take a lot of years to learn. In fact we spend most of our life learning and some never stop. But probably the most significant difference with AI is that once we have learned all that we wanted to learn, we can't make a backup of that knowledge or copy it to another human.. at least not in a way that would take significantly less time for the other human to learn it.
While you could say that once we have an AI that has learned all that we wanted it to learn, it could hold that knowledge for ever, while it continues to improve on it. Or make instant copies of it.
Of course, once you start to wonder in this direction the creep factor starts to rise. But the potential for an ever growing intelligence comes along the hand of AI.

Now if that ideal AI learns how to replicate itself with one of these:
Then we have a problem.. :)

Edit to add : This is extensively covered by the singularity subject.

2 comments:

  1. Seba, I don't think AI is going to reach true consciousnes any time this century. As Penrose says in his Emperor's New Mind, it's all crap, the boasts of these AI-ists; every 20 years we hear 'in 20 years we'll have conscious machines'. We're still not within a shout of it - see my essay for a blast against the monstrous regimen of the materialists and AI fanatics: Damn! this blog allows no cut n' paste: www.thegreatdebate.org.uk/Deasy1.html

    ReplyDelete
  2. Hugh, I respect you opinion. I'm not sure also when AI might arrive, but we are getting closer at small steps.

    ReplyDelete