[extropy-chat] RE: Singularitarian verses singularity

Samantha Atkins sjatkins at mac.com
Thu Dec 22 05:42:30 UTC 2005


Sigh.  A true SAI is unlikely to keep the same prejudices as the  
programmer[s] that wrote the initial code.  This has been covered  
countless times.   The idea that is likely to be most realistic  is  
to write a "seed" AI that is powerful enough to recursively optimize  
itself.  Only rare programmers will be able to create or understand  
the seed.  No human is likely to understand the code after one much  
less several optimization passes.

As much as I like Open Source it is likely to be not terribly  
relevant or useful except for the human design and optimization of  
subsystems for the seed.

- samantha

On Dec 21, 2005, at 5:44 PM, gts wrote:

> On Wed, 21 Dec 2005 20:15:48 -0500, The Avantguardian  
> <avantguardian2020 at yahoo.com> wrote:
>
>> Obviously. Since anyone who programs an SAI will tend
>> do so to favor the "tribe" that the programmer belongs
>> to, nobody will trust the motivations of an SAI that
>> was built by a different tribe even in the unlikely
>> chance that the programmer was such a rare altruistic
>> soul to actually design the SAI to be "tribe-neutral".
>
> I would hope they would be open-source, especially if they will be  
> making public policy decisions.
>
> Of course the skill to understand the code would be rare or non- 
> existent if SAI's start writing code for better SAI's.
>
> -gts
>
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo/extropy-chat




More information about the extropy-chat mailing list