8 comments

[ 3.2 ms ] story [ 28.6 ms ] thread
There is no evidence at this point that we're even close to having an AGI.
Indeed. If Kurzweil and Schmidhuber want to predict the future by extrapolating a curve, they should try the one for the amount of AGI technology we have, which so far is a flat line at zero.
Deepmind's alpha go is general ai.

Have you been under some strange rock absent internet access?

As far as I know when we say AI today we just mean computers that perform various mathematical calculations in a way that makes it appear superficially like human reasoning. That is no where near the kind of cognition that humans or animals do and there is no reason to think we have even touched or started on machines that can do the type of cognition that humans engage in.

Computation is not cognition.

appear superficially like human reasoning

Exactly.

Anyone who wants to see the perfect example of this should watch Watson totally blow the Final Jeopardy question by not understanding what is the meaning of "U.S. cities". http://techland.time.com/2011/02/16/why-did-watson-think-tor...

Watson is a fabulous tool, but it's nothing like "the kind of cognition that humans or animals do".

One of the biggest reasons Watson is able to beat the champions so thoroughly is that it has a huge advantage in buzzer reaction timing. If that element were taken out, it would be much closer.
or somewhere in between.