[ExI] How not to make a thought experiment
jonkc at bellsouth.net
Thu Feb 25 16:08:09 UTC 2010
Since my last post Gordon Swobe has posted 4 times.
> I define weak AI as capable of passing the Turing test but having no subjective mental states.
Another good definition of "weak AI" is the only type of intelligence Evolution could produce if consciousness were not linked with intelligence.
> Strong AI also passes the TT, of course, and has subjective states.
And strong AI CANNOT be produced by Evolution unless Swobe is wrong.
> Why should an unconscious weak AI brain follow the architecture of a conscious biological brain? It seems you want to burden AI researchers with a pointless constraint.
It seems you want to burden Evolution with a pointless constraint.
> You claim to deny the possibility of weak AI, or claim that weak AI = strong AI. That seems to me incoherent.
Lots of things would be ignorant to someone like Swobe who is completely innocent of any knowledge about how Darwin's Theory of Evolution actually works.
> you cannot get identical external behavior from an unconscious component.
And yet in his next breath Swobe assures us the Turing Test cannot provide even a hint of underlying consciousness; as I said before consistency is not Swobe's strong point.
> Unlike observer-independent objects like mountains and planets, computations are always observer-relative. They exist only relative to the mind of some observer who does the computations.
So to observer A 2+2 =4, but to observer B it is 5, and to observer C it is 3, and to observer D it is both 3 and 5 but never 4. Does anybody else thing that is dumb?
I'd like to say something about Swobe's unique debating style; ignore any substantial criticisms brought up and hope people will just forget about them, instead just keep saying the same tired old arguments over and over and hope pure repetition will eventually wear down your opponent. I didn't find Swobe convincing two and a half months and many hundreds of posts ago when he started all this, and today I find that Swobe's ideas age like a fine milk. I would humbly suggest that it might make for more interesting reading if rather than go on and on about a malevolent person "who goes by the name of John K Clark" he might try actually defending himself against the points raised. I don't claim that this radical change in tactics would convince more people to his way of thinking, but it would make Swobe's posts more interesting.
John K Clark
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the extropy-chat