I agree, it doesn't require technical AI skills but I would expect a person in this position to have SOMETHING relevant in their background that they could point to as a qualifier.
Honestly, I read the article and it left me feeling quite sympathetic to this lady. It seemed like another clear cut case of political correctness run amok. Then I looked up her bio online and found she has a Bachelors degree in Education from the Hampton Institute and her entire career has consisted of a series of political appointments the highlights of which include Director of the Office of Personnel Management (despite not having any HR related background) and the Virginia Secretary of Health and Human Services (despite not having any public health background). She has neither training nor experience in either AI or Ethics. Politics aside, she was not even remotely qualified to be on an AI Ethics Board for Google.
Many successful teams in history included members who were not technically skilled, but had an implementable vision. The gap in knowledge bridgeable, and the alternative is more "Facebook is involuntarily committing users to an experiment in what makes people sad" ethical errors.
Tech people can design the tech. But here we are talking about the impact and those people should not be the one deciding everything. They are doing that over a decade and disappointed.
Tech companies are not made only of tech people. Those companies you named employ many engineers but are led, like any other company, by finance, sales and management. This whole story is actually about a conflict between the rank engineers and the management (similar to the earlier conflict about military projects).
You should totally trust the tech people, as they are the ones who are not in for the money (or less likely to be just in for the money, at least).
Actually, the whole story of Google not being trustworthy any longer might just be the story of engineering being slowly overruled by finance/management there.
I would totally trust Google AI researchers, like i would totally trust Einstein with the fussion theory. That employees were still able to overthrow a comitee formed by upper management will achieve more than whatever little benefit this comitee would have achieved.
I mean, human drivers' education doesn't cover choosing who to kill in unavoidable crashes. Isn't that because we believe crashes where the driver can't avoid the crash, but can choose who to kill, are so rare as to be negligible?
IMHO much more realistic and pressing AI ethics questions surround e.g. neural networks for setting insurance prices, and whether they can be shown not to discriminate against protected groups.
The main focus of "AI ethics" needs to be on model bias and how to counter it through transparency and governance. More and more decisions, from mortgage applications to job applications are being automated based on the output of some machine learning model. The person being "scored" has no insight into how they were scored, often has no recourse to appeal the decision, and in many cases isn't even aware that they were scored by a model in the first place. THIS is what AI Ethics needs to focus on, not navel gazing about who self-driving cars should choose to kill or how to implement kill switches for runaway robots.
I'd look at a few other reasons:
- We don't have "driving ethics" classes at all. Human driving education covers how to drive. "AI ethics" might cover many things, but I don't think "how to drive" is on that list. That topic falls under "AI", not "AI ethics".
- The usual example you hear about is an AI driver choosing whether to kill a pedestrian or the driver. There is no point in having a "driving ethics for humans" class which teaches that it is your moral duty to kill yourself in order to avoid killing a pedestrian. No one would pay attention to that, and people would rightly attack the class itself as being a moral abomination.
This example actually makes me more sympathetic to the traditional view that (e.g. for Catholics) suicide is a mortal sin, or (for legalists) suicide should be illegal. This has perverse consequences, like assuring the grieving family that at least their loved one is burning in hell, or subjecting failed suicides to legal penalties. But as a cultural practice, it immunizes everyone against those who would tell them to kill themselves.
To anyone who is an expert, this is a profoundly uninteresting question. Literally no modern system is programmed this way, and many people would argue that telling a system who to hit is, itself, unethical.
A more interesting question might be if our models will hit certain groups of people more often, without anyone having explicitly asked them to.
Archive link to the short bios, since Google seems to have taken the document down: https://web.archive.org/web/20190331195013/https://ai.google...
Just because you know a thing or two about ethics doesn't mean you are equipped to discuss a particular domain.
How can you possibly begin to discuss AI Ethics if you have no idea how it works or what's -realistically- possible?
What matters for ethics is the effects things have on people. You generally don't have to know how something works in order to understand what effects it has--someone who does understand how it works can figure that out and tell you.
For example, if you had to decide between on the ethics of using a nuclear weapon and using conventional weapons to destroy some legitimate military target you wouldn't need to know anything about the physics of nuclear weapons.
All you'd need to know about the nuke is how powerful it is, the long term illnesses it can cause, how it can affect people far away from the blast, how it can make the area unsafe for humans for a long time, and so on. To decide the ethical issues you are concerned with what happens, not how it happens.
If we ever get to general AI, and are dealing with ethical questions like whether it is murder to restore a robot to factory settings, whether it is slavery to own a robot, or whether a robot can own property then we will probably need ethicists who are also AI experts.
In this case “using a nuclear weapon” is easier to reason about for a non expert. What about “using nuclear technology for renewable energy”? If the person doesn’t really understand the pros and cons of this by nature of being a domain expert, they’re just relying on whatever information they may have (incorrectly) learned or been indoctrinated about.
Otherwise smart ethics people may make stupid decisions because they think they understand what they’re talking about, but actually do not.
Just take existing domain experts and train them in ethics.
How about pushing this to congress. None of them know hardly anything about anything. They delegate a lot of their thinking.
So two problems.
1. People who are practitioners are more likely to be for the technology than against it. Tristan Harris is a good example of what you're looking for. 2. Going to the logical extreme on this doesn't work.
P.S. Should we apply this to journalism? Because seriously journalists these days don't even know how to make phone calls to even pretend to fact check.
I don’t buy your argument that all experts are necessarily proponents. Even within a domain there are disagreements.
The government issue is real and also slightly tangential. We need to make working in the public sector more attractive.
And yes it should apply to journalism, but discerning that falls onto individual people. It’s a bit of a different issue.
Well then we disagree. Being an engineer or a technician does not make you a good ethicist. And that's what we need.
Training an ethicist that is impartial or thoughtful from the beginning about the technology may also be easier than the opposite. They may be similar...
But training an engineer in ethics I think is a good step. Some fields, like medicine, have it somewhat built in. We can debate how effective or serious that actually is.
Being a technician or engineer does predispose you to thinking what you are working on or working with is ethical. I did list Tristan Harris as a good counter-example and someone that certainly can speak to the ethics of the issue. But his example is also a good example of engineers/technicians not being good candidates for being impartial because he has to be a type of activist.
> I don’t buy your argument that all experts are necessarily proponents. Even within a domain there are disagreements.
I said likely, not exclusively.
That's what other members of the board are there for. Let's flip your question: How do you expect a tech person to comment reasonably about AI ethics when all they've taken is an undergrad course in philosophy (and even that may be a stretch).
The notion that a person on the board must be an expert in every aspect involved is ridiculous.
I wouldn’t want John Stuart Mill on that board because he wouldn’t know what he was talking about, and therefore would be unable to properly evaluate things.
I'm getting the sense that the goalposts are ready to slide here, but I wouldn't expect an expert on "when to not do a new Holocaust" to be able to model interactions between uranium atoms.
Not having any familiarity with the subject is obviously its own disqualification as an expert in a given area of ethics. And a nuclear physicist could lack understanding of ethics, international politics, history, military, who's even armed... This would be far more disqualifying.
In this case I don’t care how trained an ethics person is in ethics or history. They literally are not equipped to discuss this topic without being fed information from somewhere, which leads to its own issues.
I would much rather take domain experts and train them in philosophy, ethics, and history to an extent. That is far easier and better than the other way around.
Publication records help to separate the wheat from the chaff.