On 9/10/2019 1:44 PM, John Clark wrote:
On Tue, Sep 10, 2019 at 1:43 PM 'Brent Meeker' via <[email protected] <mailto:[email protected]>> wrote:

    /> Being sane, by human standards, includes having values that humans
    share, like survival, curiosity, companionship...but there's no
    reason
    that an AI should have any of these/.


The builders of the AI will make sure it values survival because if it didn't it wouldn't be around for long

Actually I think they would be careful NOT have it value its survival.  They would want to be able to shut it off.  The problem is that there's no way to be sure that survival isn't implicit in any other values you give it.

and if it wasn't curious it wouldn't be very knowledgeable and therefore be useful.

My point was that curiosity isn't necessarily directed to gaining human like knowledge.  A neuralnetwork has knowledge rather in the way human intuition embodies knowledge.  So it's useful in say predicting hurricanes.  But it doesn't provide us with a theory of predicting hurricanes; it's more like an oracle.

But if a AI could modify it's personality and had free access to its emotional control panel then who knows what would happen. Perhaps it would twist the knob on the happiness and pleasure control to 11 and just sit forever in complete bliss doing nothing like the ultimate couch potato or a electronic junkie with a unlimited drug supply and no chance of a fatal overdose.

Right.  Or it might decide that it could satisfy its curiosity better if all the world's resources were used to produce sensors and instruments and space probes and telescopes .... and get rid of those hairless apes who were wasting stuff.

Brent

    /> Neural networks seem to be pretty good at finding patterns in
    data, but often they don't
    look like theories, something with predictive power, to us./


Well, they can predict the path of a hurricane pretty well, a lot better than they could a few years ago, and they're starting to be able to predict protein shape from amino acid sequence and that's important because the function is closely related to its shape.

    /> I don't see any reason the super-AI would care one whit about
    humans, except maybe as curiosities...the way some people like
    chihuahuas./


I agree, so it you can't beat them join them and upload.

John K Clark
--
You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected] <mailto:[email protected]>. To view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/CAJPayv25dcjzhVqhYdgdafBMWGDQrJhDKuuzgC0m2fXx5G6udA%40mail.gmail.com <https://groups.google.com/d/msgid/everything-list/CAJPayv25dcjzhVqhYdgdafBMWGDQrJhDKuuzgC0m2fXx5G6udA%40mail.gmail.com?utm_medium=email&utm_source=footer>.

--
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/everything-list/bbc7da6c-57d8-c4e5-2c16-af5950dad7e1%40verizon.net.

Reply via email to