Ethics: Robots and AI

Post Reply New Topic

Feel welcome to use this thread for any ideas you have on the topic.

Here's mine:
Is it ethical to do whatever you want to a robot? Where do you draw the line? Does it matter if the robot is humanoid?

Is it ethical to do whatever you want to an AI?

No interest in discussing sex robots? combat robots?

I wouldn't do anything to a robot that would make me feel bad, or that I would not want to see on a video shown to me by scary people. I don't know what the ethical boundaries are, exactly, but I feel 100% certain that they are out there.

I think a reasonable starting point is to propose that if a thing is so like a living being that a typical person reacts to it as if it were such being, human or animal, then treating it ways that would be abusive if it were truly such a being are morally equivalent to doing those things to a living human or animal being.

Sam Harris and Paul Bloom had an interesting dialog around this topic in relation to the film "ex machina" and TV series "Westworld" that might be interesting to you:

Personally, I'm curious about how shutting off an AI would work. Would it count as killing one, or simply making it sleep? Would it put it in a "temporarily dead" situation, where restarting the program would revive it, or would it just create a new one with the old one's memories and personality?

Also, patching a program requires it to be shut off, so if turning it off counts as killing one, it'd be impossible to ever update or make changes to it.

I like the video, Adamantium. They make excellent point after excellent point. Regarding what I said in my answer, we seem to have the same idea. It isn't about the machine itself, it's about how we relate to something else. Creating empathy with a doll and then abusing it diminishes people, but it gets worse if the doll has some capacity to suffer.

Their point about the feasibility of owning superintelligent self-aware slaves is a good one. Cylons.

They did not explore one other point to my satisfaction. I'd like to stipulate that a specific feedback mechanism qualifies as pain. The Roomba might not be the best example, but I'll roll with it. It isn't designed to be intelligent, but it has lots of equipment to map, one could even say sense, it's environment. If something in the environment starts to damage it, this feedback mechanism is activated, warnings go off, it moves away from the stimulus, one could say pain. It's got math processors to deal with all this, and complex software to prioritize tasks, so it isn't dumb. One of it's jobs is to stay out of the owner's way.

So, the Roomba Mark512, for whatever reason, keeps getting in the owner's way. The owner takes it personally, and smashes the Roomba with a hammer. Not enough to completely disable it, but it's clearly limping around with alarms going off, still trying to execute it's program. Is that cruel?

Just to be clear, if we change only the physical form of the Roomba so that it is a convincing humanoid with essentially the same hardware and software, hitting it with a hammer would be cruel?

I don't find that satisfying.

My dissatisfaction with your limit raises another question, on which I assume Philosophy has settled on an answer. What do we mean by ethics?

I tend to regard ethics as the individual expression of morality, the collective social expressions of ethics. The Ten Commandments are morality for some, although we disagree violently about number one. Don't kill people is a widely held moral view, although a person can have that view of ethics without accepting one particular capricious and lying god's alleged moral code.

So, where will we draw the line on the morality side? Will we ban some or all sex robots?

The fighting robots matter, too. I contrast Robot Wars, one of my favorite shows, with the scene in Mrs. Peregrine of the strange fighting creatures which made me feel overwhelming nausea, and not just from looking at them. That scene was clearly designed to give the audience a negative impression of that character.

I'd say your point about the biblical distinction between killing and murder, though accurate and revealing, demonstrates that morals evolve.

So we are already regulating that kind of behavior. See the thread BBC Bans Bunny Game, although fair warning you can't unsee it once you do.

Do we imprison people? Strictly regulate the types of robots which can be produced? People go to prison for drawings in various places around the world, or get assassinated over them. It seems likely that we will just adapt that system to include this new technology. That's what we usually do with new tech, and it sort of works for a little while.

Would the very making of a Mohammed, peace be upon him, robot be unethical?

I see it as a whole basket of ethical concerns. What about the sex robots? Their very existence will shock the conscience of some, not to mention their uses. I predict that they will be regulated, but I don't see how. Will people talk about what they do with their robots in the privacy of their own homes being their own business?

For that matter, do we make harming a robot illegal, and if so at what point?

I think the more powerful our AI gets, and I do believe it will be dumb-AI for a very long time perhaps our entire lifetimes, keeping them highly specialized will be critical for managing their capabilities.

Some guys at work were making a joke about what would happen if you told a very powerful and universalized dumb-AI to make people happy; that it would quite likely put human beings in stasis with some kind of dopamine and serotonin drip or, voting that being alive causes more misery than not, would kill off humanity with heroin overdoses.

The biggest concern seems to be runaway effects. This is part of why you don't want to universalize a dumb-AI's capabilities.

As for representations and their abuse - I think this one of those areas were most people won't even believe its a thing until it starts happening and it may be a very short-lived issue or it may draw intense social stigma, which would be appropriate.

From my reading of history, it seems there have always been people who would f**k a snake if they could hold it still, and I see no reason to expect that they will ever cease to exist. So, I don't know how any new way to use tech for sex will die out. I consider it plausible that a few erotic daguerrotypes are still produced each year.

Society will not be satisfied with prosecuting people who create images using sex robots; they will not tolerate people doing creepy things at all. The very existence of realistic child sex robots will fill enough people with murderous rage that a moral panic seems inevitable.

I keep contrasting the sex robots with the combat robots because I find society's views of those interesting to contrast with each other. For example, Robot Wars will eventually have semi-autonomous robots unless the rules prohibit it. When will people start to worry about creating AI with no purpose but to destroy each other? I think it will be later than for the sex robots, because moral panics rarely arise against non-sexual violence.