[ExI] Limiting factors of intelligence explosion speeds

[Eugen Leitl]

On Fri, Jan 21, 2011 at 07:21:18AM -0500, Richard Loosemore wrote:
> Eugen Leitl wrote:
>
>> Speaking about stupid ideas, you're pretty good at these.
>
> Address the topic.

Judging from the strawmen, you're having issues with fitness
function for agents. Of course you can define more or less
sensible synthetic fitness functions, but in practice if you're
bootstrapping useful parameters from a large space, say, for
neural control you're having applications in mind, or applications
will emerge as soon as the system does something interesting.

What that is, depends on purpose. You could use it to solve
CAPTCHAs, compromise systems for phishing or use it for trading
or use it for weapon platforms control, or ability to derive 
ad hoc languages for hunt pack or construction crew communication.

And of course there's an intrinsic fitness, where usability of
a platform is being pitted against competition. Does it crack
trade or kill better than the others? Then it will be used more.
And it's not your decision, but the decision of the human 
population acting as a whole. Eventually, as autonomy increases
and tools become persons the impact of human decisions shrinks
and eventually becomes completely irrelevant.

-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE