[ExI] Oxford scientists edge toward quantum PC with 10b qubits.

Richard Loosemore rpwl at lightlink.com
Wed Feb 2 16:56:30 UTC 2011


Kelly Anderson wrote:
> On Mon, Jan 31, 2011 at 1:42 PM, Richard Loosemore <rpwl at lightlink.com> wrote:
>> spike wrote:
>> Watson does not contain the germ of an intelligence, it contains a dead-end
>> algorithm designed to impress the gullible.  That strategy has been the
>> definition of "artificial intelligence" for the last thirty or forty years,
>> at least.
>>
>> A real AI is not Watson + extra machinery to close the gap to a full
>> conversational machine.  Instead, a real AI involves throwing away Watson,
>> starting from scratch, and doing the whole thing in a completely different
>> way ....  a way that actually allows the system to build its own knowledge,
>> and use that knowledge in an ever-expanding range of ways.
> 
> Richard, Is your basic problem with Watson that it is going in the
> wrong direction if the eventual goal is AGI? Are you concerned that
> the public is being misled into believing that computers are closer to
> being "intelligent" than they actually are?
> 
> I'm trying to understand the core of your indignance.

Well, both, and more.

People were complaining about this kind of cheap-trick AI at least two 
decades ago, and we expected that our complaints would be loud enough 
that it would eventually stop.  But it did not.  Every few months, it 
seems, there is another announcement about some project, which the press 
writes up as "Could it be that AI is on the brink of a breakthrough?". 
Can you imagine how indignant you would be if you saw those same stories 
being written 20 years ago?  :-)

I guess one of the reasons I am personally so frustrated by these 
projects is that I am trying to get enough funding to make what I 
consider to be real progress in the field, but doing that is almost 
impossible.  Meanwhile, if I had had the resources of the Watson project 
a decade ago, we might be talking with real (and safe) AGI systems right 
now.


Richard Loosemore



More information about the extropy-chat mailing list