[extropy-chat] AI and the law already

[Anders Sandberg]

gts wrote:
>> but if AIs now need a license...
>
> They don't.
>
> I was going to comment about what a strange and remarkable legal decision
> this seems to be, but then I read the article and realized what should
> have been obvious: contrary to the headline and the words in the first
> paragraph, the judge did not actually rule that "the software was
> effectively practicing law without a license". The ruling was against the
> human seller of the computerized legal service, of course.

Which to some extent is a ruling against the human+AI system. Only the
human part has a legal standing, but it is the absence of a accredited
lawyer part of the system that is the legal problem. One might imagine a
future version where there is indeed a lawyer in the system, but most of
the interaction is going on anyway through the AI. At that point I guess
the law has to become more firm on how tight the coupling has to be
between the lawyer and the AI;  likely it would not be enough that the AI
was doing all the work and the lawyer was just sipping tea all day.

AI rights are much simpler than rights and responsibilities for this kind
of composite systems. But if we can get them right, maybe it is possible
to apply the same kinds of reasoning to densely connected by composite
systems like human minds.

-- 
Anders Sandberg,
Oxford Uehiro Centre for Practical Ethics
Philosophy Faculty of Oxford University