[ExI] The fraudulent claims made by IBM about Watson and AI

[Rafal Smigrodzki]

On Mon, May 30, 2016 at 10:36 PM, John Clark <johnkclark at gmail.com> wrote:
>
>
> ​Granted computers aren't smart ​about everything, but then people aren't
> either. But c
> omputers already show greater than human intelligence in playing Chess and
> Go and ​Jeopardy and in finding the websites that tell you what you want to
> know. And if
> "​
> actual insight
> ​" isn't needed to do that then
> "​
> actual insight
> ​" isn't very important.
>
> ### I'd rather say, actual insight is present, in some way, whenever any
question is answered with above-random precision. It may be just echoes of
the real world encoded in word frequencies and word associations but still
there is insight. An emmett knows her queen's desires. A self-driving car
or a legged robot have insight into the physics of moving objects, however
limited. A bean stalk knows light from darkness. Watson has an inchoate
understanding of the world, however unimaginative. Of course, the go
program has insight into what moves win in go.

One might claim that "true" insight is given only to a device that holds a
quantitative and predictive true model of everything in the world that can
be modeled. No human has ever had anything close to that godlike power and
no AI ever will. Still, even a tiny amount of insight, a most meagre
predictive model of the world, are crumbs of truth.

Rafał
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20160531/a073a5d9/attachment.html>