Re: A Godfather of AI Just Won a Nobel. He Has Been Warning the Machines Could Take Over the World.

On 2024-10-12, kazu <f00@0f0.00f> wrote:

Borax Man wrote:

["Followup-To:" header set to comp.os.linux.advocacy.]
On 2024-10-12, kazu <f00@0f0.00f> wrote:

John Smyth wrote:

'A Godfather of AI Just Won a Nobel. He Has Been Warning the Machines
Could Take Over the World.'
>
<https://archive.is/VuJ4L#selection-2403.0-2597.172>
>
>
'The newly minted Nobel laureate Geoffrey Hinton has a message about the
artificial-intelligence systems he helped create: get more serious about
safety or they could endanger humanity.

>
>
i am 50-50 on the view that AI will pose a danger to humanity.
>
both sides have decent arguments and i cant make up my mind.

 
I don't think AI will take over, as it is not really capable of
out-of-band thinking.  AI cannot solve a problem which hasn't been
defined.  For example, you can get AI to perhaps streamline a process or
design, make it more efficient, but it won't come to the realisation
that the thing you are trying to do, may not be necessary after all.  So
AI would have to do something it is not capable of at the moment, coming
to new realisations and acting on that information.
 
The threat from AI I think is more how we act.  I see a danger where we
trust AI, and treat it like it is a superintelligence, and we use it as
a crutch to make desicions, which turn out to be poor, suboptimal
decisions.
 

>
current models may not be able to, but there is an exponential
curve we are riding.

What we have seen is better models because they have more data.  But
there is a leap to be made to get them to think out of band.  This is a
very different type of program, and now where I think the A.I. path is
leading to at the moment.

Machines can't think, and programs can't think.