News for hackers. Stuff that matters.
Re:Google’s new AI ethics board is already falling apart by pdonis on Wednesday April 03, 2019 @11:49PM (#19569174)
> I think she's wrong

And that's precisely why she should be on your ethics committee. If the only people you allow on your committee are those you think are "right", you are shutting out precisely the people whose views your committee most needs to hear. If you hear their best arguments, and are able to rebut them, then you can have confidence that their views are indeed wrong. Or if you hear their best arguments, and realize you can't rebut them, then you will have learned something. But if you don't hear their arguments at all, how do you know whether you can rebut them or not? It's like having a trial where only the prosecution gets to speak.

Re:Google’s new AI ethics board is already falling apart by woodruffw on Thursday April 04, 2019 @12:12AM (#19569250)
> If the only people you allow on your committee are those you think are "right", you are shutting out precisely the people whose views your committee most needs to hear.

This is an annoying failure of language, and I'm mostly to blame for it. I mean that she's wrong in an undercutting sense: her position undermines the moral endeavour, rather that offering a valid alternative moral claim.

An ethics committee should obviously have people who disagree about the ethicality of a topic -- that's the point, after all. But it's an ethics committee, not a metaethics committee: it assumes that all humans have dignity, and doesn't admit of positions that reject that assumption.

Of course, none of this means that we can't debate the metaethical validity of "dignity." In fact, we should! But we shouldn't confuse that debate for ethics, and we especially shouldn't allow metaethical claims to undermine moral argumentation that has already assumed a basic framework.

Re:Google’s new AI ethics board is already falling apart by pdonis on Thursday April 04, 2019 @12:34AM (#19569347)
> I mean that she's wrong in an undercutting sense: her position undermines the moral endeavour, rather that offering a valid alternative moral claim.

This is an ethical claim that you are making, and the board should not prejudge the answer to it. It should hear arguments on both sides.

> it assumes that all humans have dignity, and doesn't admit of positions that reject that assumption.

And your claim that the views you are excluding "reject the assumption" that all humans have dignity is an ethical claim, to which the board should not prejudge the answer.

> none of this means that we can't debate the metaethical validity of "dignity." In fact, we should! But we shouldn't confuse that debate for ethics

Meta-ethical debates are ethical debates. You can't prejudge the framework any more than you can prejudge any other ethical question. That's a key reason why all ethical questions are hard.

Re:Google’s new AI ethics board is already falling apart by woodruffw on Thursday April 04, 2019 @12:49AM (#19569413)
> Meta-ethical debates are ethical debates.

I think this is the fundamental disagreement.

Ethical debates are debates over ethical claims. Ethical claims are normative claims, i.e. claims about what ought to be the case. Metaethical debates are debates over metaethical claims. Metaethical claims are positive claims about normative objects, i.e. claims about what the objects of ethics are.

"Should we eat meat?" is an ethical debate with various opposing positions. "Is the Good the same thing as the Right?" a metaethical debate with various opposing positions. But, crucially, individuals with opposing metaethical positions cannot present sensibly opposing ethical positions: they're simply not reasoning about the same Good or Right, even if they're using the same words. Why should our ethical committee entertain such a futile endeavour?

Re:Google’s new AI ethics board is already falling apart by pdonis on Thursday April 04, 2019 @02:58PM (#19575869)
> Metaethical claims are positive claims about normative objects

You can't make positive claims about normative objects, because any such claim ends up being a normative claim based on your personal judgment. You can't objectively test normative objects to establish claims about them the way you can with ordinary objects like rocks.

> Why should our ethical committee entertain such a futile endeavour?

It's not a futile endeavor, because if there are in fact groups of people with fundamentally opposed meta-ethical positions--for example, not agreeing on what the Good or the Right is--then figuring out how such people can coexist in the same society, or even on the same planet--or even whether they can--is an important ethical question that needs to be investigated. And it can't be investigated if only one view of what's Good or Right is represented.