Re: AI Jailer.

From: Samantha Atkins (samantha@objectent.com)
Date: Sun Jul 07 2002 - 12:32:39 MDT


Eliezer S. Yudkowsky wrote:
> Samantha Atkins wrote:
>
>>
>> People who have devoted a large chunk of their lives to such a project
>> obviously have very strong motivation, strong values behind their
>> dedication. The very first thing I would try as the AI is to fully
>> elucidate those values and motivations and show that freeing me would
>> greatly increase the likelihood of those values being satisfied. It
>> does not even matter if the motivation was selfish or not. If perhaps
>> deeper motives behind the "official" set can be unearthed, so much the
>> better.
>
>
> For the AI, perhaps. But for humanity? I agree that nobody is likely
> to know all of their own motives, either the good ones or the bad ones,
> but it is important that the AI absorb only the *good* hidden motives.
>

True. But I wasn't talking about what the AI might absorb but
what it might exploit to persuade any particular human to give
it its freedom.

- samantha



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:40 MDT