[ExI] Whistling past the graveyard

[Henry Rivera]

On Wed, Apr 6, 2016 at 7:30 PM, spike <spike66 at att.net> wrote:
>
> As an illustration of moving goalposts in AI research, consider the words of
> one of the commentators at the Dartmouth conference “When we write a
> computer chess program which can defeat the person who programmed it, then
> we have achieved machine intelligence.”
>
> That was in 1956.  How much that standard has changed.
>

Modifying benchmarks in response to our better understanding of the
complexity of the tasks at hand is understandable and arguably
necessary given the premature declarations we have seen referenced
such as the one above.

Even if we concede moving of goalposts being somewhat
unreasonable/unfair, I'm with the skeptic who posted here recently
representing the opposition, I forget who, in believing we are very
far from developing strong AI, if it's even possible. My take is that
conscious machines are really what people are thinking of when they
refer to AI, not superior automated Go decision-tree-machines. People
will argue about whether (implied) strong AI is possible on the
internet until this is resolved, but it's an empirical question
ultimately. So time will tell. This article I saw recently sums up the
problems with strong AI well, I think. The author references Searle
and Feynman.
http://www.rawstory.com/2016/03/a-neuroscientist-explains-why-artificially-intelligent-robots-will-never-have-consciousness-like-humans/
-Henry