Re: When does it pay to play (lottery)?

From: Brian Atkins (
Date: Sun Jan 23 2005 - 15:00:43 MST

Ben Goertzel wrote:
>>-----Original Message-----
>>From: []On Behalf Of Brian
>>Sent: Sunday, January 23, 2005 1:35 PM
>>Subject: Re: When does it pay to play (lottery)?
>>David Clark wrote:
>>>The chance that *any* implementation of AI will
>>>*take off* in the near future is absolutely zero. We haven't
>>the foggiest
>>>clue exactly what it will take to make a human level AI, let alone an AI
>>>capable of doing serious harm to the human race.
>>Your two statements of certainty are in direct conflict with each other,
>>so I don't see how you can hold both at the same time.
> I don't see that his attitude is self-contradictory.
> I would say "The chance that anyone will build a ladder to Andromeda
> tomorrow is effectively zero. We don't really have any clear idea what
> technology it would take to build such a ladder." No contradiction there.
> I don't know how to do it, but I know that none of our current methods come
> close to sufficing.

Exactly - you know because you know (or can fairly easily find out) what
the actual structural requirements are that such a ladder would have to
meet in order to succeed. This is a very different situation than AGI
knowledge is at currently. So your analogy is a very poor one in my opinion.

David was making statements about a technology where we still don't know
what the exact "structural requirements" are for it to work. For a
ladder we can easily use existing physics knowledge to determine whether
a design will hold up or collapse, and we can see for relatively certain
that we don't have the techniques to make it work, but for a given AGI
design there are no such hard and fast rules of such precision and
predictive strength.

Brian Atkins
Singularity Institute for Artificial Intelligence

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:50 MDT