[extropy-chat] The emergence of AI

[Keith Henson]

At 08:31 PM 03/12/04 -0500, Eliezer wrote:
>Hal Finney wrote:

snip

>>Are you serious?  30 seconds, once the AI reaches human level?  What on
>>earth could yet another human-level contributor to the team accomplish
>>in that time?
>
>"Human level" is an idiosyncratic flavor.  I am talking about an AI that 
>is passing through "roughly humanish breadth of generally applicable 
>intelligence if not to humanish things".  At this time I expect the AI to 
>be smack dab in the middle of the hard takeoff, and already writing code 
>at AI timescales.

I have a strong emotional inclination to support Hal's view rather than 
Eliezer's apocalyptic view.

However . . . .

About two years ago there was that virus, I forget the name, that infected 
something like 75,000 vulnerable Microsoft database machines that were 
connected to the Internet.  Made a mess of the net for a few days till the 
machines were taken off line and patched.

But what's important about this episode and possibly provides an 
instructive real world example for AI timing is that the number of infected 
machines was later analyzed from the records to have a doubling time of 8.5 
plus or minus one second.  People just can't react to a threat that comes 
"out of the blue" this fast.

If an AI was smart enough to write a virus to take over machines and its 
intelligence was proportional to "cortical area" (processor cycles) and it 
"wanted" to get smart fast . . . . .

:-(

Keith Henson