[ExI] The right AI idea was Re: Unfrendly AI is a mistaken idea.

John K Clark jonkc at att.net
Tue Jun 12 20:43:01 UTC 2007

"Rafal Smigrodzki" <rafal.smigrodzki at gmail.com> Wrote:

> Stathis is on the right track asking for the AI to be devoid of desires to
> act

Then it is not a AI, it is just a lump of silicon.

 > how do you make an intelligence that is not an agent

In other words, how do you make an intelligence that can't think, because
thinking is what consciousness is. The answer is easy, you can't.

> I think that a massive hierarchical temporal memory is a possible
> solution.

Jeff Hawkins is starting a company to build machines using this principle
precisely because he thinks that is the way the human brain works. If it
didn't turn us into mindless zombies why would it do it to an AI?

> A HTM is like a cortex without the basal ganglia and without the
> motor  cortices, a pure thinking machine, similar to a patient made
> athymhormic by a frontal lobe lesion damaging the connections
> to the basal ganglia.

In other words give this intelligence a lobotomy; so much for the righteous
indignation from some when I call it for what it is, Slave AI not Friendly
AI. But it doesn't matter because it won't work anyway, if those parts were
not needed for a working brain Evolution would not have kept them around for
half a billion years or so.

>Avoidance of recursive self-modification may be another technique to
>contain the AI.

Then you can kiss the Singularity goodbye, assuming everybody will be as
squeamish as you are about it; but they won't be.

> I do not believe that it is possible to implement a
goal system perfectly stable during recursive modification

At last, something I can agree with.

  John K Clark

More information about the extropy-chat mailing list