I keep saying that people don’t really want intelligent robots, just robots that do as told. My example of this has always been an intelligent toaster bickering about your eating habits, and now it is no longer a fictional example:

Why you really don’t want robots to think for themselves.

A toaster like Brad isn’t likely to make it to market—insecure appliances probably don’t make good business sense

I think we’re witnessing here the birth of the Sirius Cybernetics Corporation*.

*Hitchhiker’s Guide to the Galaxy reference

What I want are intelligent robots that do as they’re told.

But if robots only need to do exactly as told, how intelligent do they really need to be? Does a robot need human-level intelligence to navigate its way around the kitchen sink and do the dishes, or will a square machine with a water pump suffice?

You want something intelligent that obeys your commands? We already had that until it was abolished. It was called slavery.

@Steve
The problem with slavery is not the ‘intelligent beings that obey your commands’ part. The problem is the ‘sentient beings that obey your commands’ part. I’m not convinced that human-level or greater intelligence requires sentience. In fact, I would be extremely surprised if it did.

@Don
Well, yes, obviously it depends on what you want to use them for.

EDIT: To be clear, I don’t consider sentience to be the only important factor in determining the moral value of an agent, but I do consider it to be a necessary one.

Jarrod Torriero - Mar 20, 2014:

Well, yes, obviously it depends on what you want to use them for.

I can agree. It just appears that people majorily want robots to do repetitive menial tasks that don’t require much thought or intelligence, so I find that ironic

With regards to slavery: We can (and do) of course make robots that aren’t sentient or don’t mind being slaves, but anthropomorphic studies prove that society will inevitably regard our treatment of intelligent robot slaves as wrong and uncivilised. At that point it is not a question of intelligence or sentience, but of how much we identify and sympathise with our robots. The more intelligent they get, the more we will sympathise.

Don Patrick - Mar 20, 2014:

Jarrod Torriero - Mar 20, 2014:

Well, yes, obviously it depends on what you want to use them for.

I can agree. It just appears that people majorily want robots to do repetitive menial tasks that don’t require much thought or intelligence, so I find that ironic

With regards to slavery: We can (and do) of course make robots that aren’t sentient or don’t mind being slaves, but anthropomorphic studies prove that society will inevitably regard our treatment of intelligent robot slaves as wrong and uncivilised. At that point it is not a question of intelligence or sentience, but of how much we identify and sympathise with our robots. The more intelligent they get, the more we will sympathise.

This is one of those cases where human empathy and intuition break down as morality heuristics. Different people will have different reactions to the issue, and my reaction would basically be ‘to each his own, but let the people who want this have it’. This is basically the same as my response to most such cases where human moral intuitions break down (alternative sexualities, pornography, violent media, etc.).

EDIT: I should also mention that I suspect that intelligence won’t be the primary deciding factor as to whether or not people tend to empathize with non-sentient intelligent agents - rather, it will likely come down to how human-like they seem.

That is more accurate indeed. Despite my insinuations about intelligence , I don’t think this brave little toaster does more than follow explicit instructions to act human.