[ExI] Whistling past the graveyard

William Flynn Wallace foozler83 at gmail.com
Thu Apr 7 23:40:59 UTC 2016


I have never understood that attitude.

I can make an evolutionary algorithm that finds solutions to problems I do
not even understand. I don't think I can take credit for that. Anders

Just how is this possible?

1 -  If you don't understand the problem how do you know you have a
solution?  What I think you mean is that the algorithm is more useful than
you thought - serendipity.  I don't think this is what you mean.

2 - Or reading your sentence another way, you know the problem, you created
an algorithm, and the answers you got you don't understand.  Then how do
you know the solution is correct?  Maybe the algorithm is flawed - that is,
it does things you did not intend.  (I am betting on 2)

I know I am out of my league here, and I have no trouble giving credit to
AI if that is due.   But the AI has to follow the programming, right?  How
does that earn credit?

If you know that the gap between what you know and what I might understand
is just too great, just tell me to shut up (politely) and I will!  (same to
John Clark)  I sincerely don't want to waste your time.

bill w

On Thu, Apr 7, 2016 at 5:43 PM, Anders Sandberg <anders at aleph.se> wrote:

> On 2016-04-07 23:30, William Flynn Wallace wrote:
>
>
> The AI may produce unexpected results, but that was known beforehand, if
> the code was OK.  So, credit to the programmers, none to the AI.
>
>
> I have never understood that attitude.
>
> I can make an evolutionary algorithm that finds solutions to problems I do
> not even understand. I don't think I can take credit for that.
>
> --
> Anders Sandberg
> Future of Humanity Institute
> Oxford Martin School
> Oxford University
>
>
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20160407/2b9301fb/attachment.html>


More information about the extropy-chat mailing list