AW: [aiethics] Let's begin...

[Christian DroЯmann]

Alicebot AI Ethics Committee - http://www.alicebot.org



> Hi All,
>
> I could understand the need for this forum to be very open,
> actually I didn't meant for it to be closed or restricted, but to
> have "something" in terms of policy to start with, sort of a
> base declaration of principles to start with, even with the
> understanding that could be reviewed at any time given the
> need to.

That's what I wanted to say with my first posting..

> Ethic systems do differ, a whole lot; key is to find out
> what the common denominators are in order to try to
> address them.

that might be for example the widely known "human rights"...

> Some a decade ago I participated on a corporate technollogy forum
> when the appointed guru predicted that by some 2030-2040
> to turn off a machine would have legal and ethical implications;

If machines have developed a kind of conscience until then, I suppose =
some
might even call it "murder"...

> But today it's a penal crime on many countries to conduct
> actions that makes machines unavailable as many virii
> related trials dictated (i.e. US vs Morris) and surely the
> charges might be hefty if the subject machine is under some
> key civil services duty. So I don't laugh anymore.

but up to now they are being charged for damaging or destroying (or
stealing) data and technical equipment, not for "hurting" a machine by
infecting it with a virus or running a DOS-attack against it...but one =
day
this might be a crime equal to hitting someone over the head with a =
baseball
bat to render him unconscious...

> One almost funny thing is that if a person makes some statements
> contained in the standard AIML set will commit
> (at least in technical law terms) a federal offense in many
> countries of what we call the Western civilization, not to
> mention a severely punished offense on some
> islamic countries. Should we address that? (please refer
> on my comments of cheap declaration of being a Global
> Village while being clueless on what it means and implies).

I think it will be impossible to find a solution that will satisfy =
everyone
on the whole planet, but I'd say we should officially disapprove of (or
to? -insert correct preposition here- ;-> )content that involves things =
like
racism or discrimination of all other kinds and other things that we =
would
all consider as "not right"...

What I meant by asking "what is NOT our purpose" was for example that we
could never use...let's say religion as a basis for our ethical values, =
as
then we would always have issues that are an offense to one religion and
perfectly alright to another...and so (IMHO) we won't be able to protect =
the
users' religious feeling in a way other than just avoiding any direct
negative statement about any religion or ethnic group in the official =
AIML
set...
And this applies to numerous things other than religion...
How we should deal with other people writing their own AIML including =
racist
content for example is a different story and should not be addressed =
before
we have developed a "ruleset" OUR alicebot must comply to...

> This nonsensical navigation is just a tiny example of the kind
> of boundless navigation this topic might encourage; and not
> working with some North might lead us to release vast amount
> of energies and still by the time the AI movie is released not
> be prepared any better than today to answer even the most
> basic questions (mention the AI movie just because Brenda
> mentioned it and it's a good example of a concrete thing
> that might require some pragmatism to be addressed).

I think pushing the discussions just for the sake of being able to =
answer
questions after the AI movie is released is not a good reason...
We should be productive in a higher sense than just having something to
show.

> Don't think for a minute it's feasible to install an agenda on
> this discussion, much less a timeline; a "master plan to
> develop an ethics strategy" sounds like a joke to me; but

This would lead itself ad absurdum...

> at least some overall statement of the things that are our
> foundation (the so called policy of the beginning) and a
> general outline of what we want to accomplish.

I agree with that...again this is what I wanted to say :-)

> One handy thing about ethical systems is that aren't usually
> enforceable (otherwise they are a law system); and I said
> it's handy because we could not possibly enforce it from
> the technical side at this point.

The other question is: If we _had_ the means to enforce it, would
"enforcing" an ethical system not ridicule itself?

> (which in the foreseable future would be of far less harm
> than the modest atomic or nuclear artifact) but in the sense
> on how the usage of a given technology might contradict
> the most inner beliefs (and thus ethics systems) of their
> creators.

When it comes to avoiding this, I'd almost say "better men than us have
tried"...

> But this would mean a nice example that once creators
> allows the genie got loose out of the bottle they could not
> control it any further; but at least we should be empowered
> to stick a tag to the genie that says that certain usages of
> it contradict the creator's set of beliefs; if not, for the peace
> of mind of the creator (and quite frankly, to start putting some
> humble legal shield on the unforeseen and undesired
> consequences of our work, the ... hmm... "lateral damage").

What about an official "seal of approval" issued by a committee at the =
ALICE
AI foundation functioning as the executive wing of the ethics committee? =
We
cannot ensure that AIML will not be used for unethical purposes =
(whatever
"unethical" may be), but at least we can make sure that a user who wants =
to
have a good clean conversation with a bot based on AIML just needs to =
look
for a bot which is A.L.I.C.E.-certified...
I'm thinking of Rich's original intention to keep Alice suitable for
children for example...
We might even extend this to ratings similar to the ones that are =
applied to
movies (G, R, X, PG...)

> Perhaps a good starting point, albeit not a novel one, are
> the Dr. Assimov "law of robotics"; since they could not be
> enforced it should be renamed by now as "ethic set of
> robotics" but behind the initial attempt to consider it a

some of them have already been adopted in the FCC rules for hardware...

Christian