Uncanny Valley dilemma

[Ihe]

For those who don't know the term, the uncanny valley refers to the sense of eeriness and revulsion people often feel when seeing humanoid things that appear almost, but not exactly, human.

With that out of the way: I've been working on a story idea for a while now, and one of the plot devices that would make my life easier is to have artificial assistants who look almost the same as their owners, with practically the same capabilities (but not quite, to avoid confusion). So, I'm looking for reasonable justifications for this to be so, despite the uncanny valley problem. Any ideas on this?

 

[Toby Frost]

I've been thinking about this quite a lot of late. Justifying building an android of this sort is quite hard: I can see why you'd make a human-shaped robot, but the effort required to make one that looks entirely lifelike would be extreme (and a bit pointless if sensors could detect it). Also, why not just make a friendly-looking machine, like Wall-E? It would certainly make repairs and disposal a lot less disturbing.

The best answer I can think of is that these robot doubles are not designed for close interaction, slightly malfunctional or not quite "finished" yet somehow. It might be that they need a period of socialising before they can get the hang of dealing with humans, and these ones aren't quite broken in yet. Alternatively, they might just find it difficult full stop: a fair amount of human interaction relies on nuance and knowing what to do in given situations. If they can't pick up that sort of thing, they might seem awkward and unsettling.

The only thing I can't quite get is that they would look like their owners. I think I'd find a robot version of me very weird. I would expect an android (at least a friendly one) to be quite mild and uncharismatic; likeable but not attractive (as to whether that's like me, no comment!). Perhaps an entirely different society might regard having doubles as a status symbol, or a sort of amusing pastime.

 

[Ihe]

The only thing I can't quite get is that they would look like their owners. I think I'd find a robot version of me very weird.

Sorry, I meant owners in the general term--human. And far into the future. I should've mentioned that.

These droids would be at least as capable as a human. Looks AND capabilities. They wouldn't be exactly the same--in order to distinguish them from humans--but the AI at that point in time would be very advanced, so social performance wouldn't be too limited.

And my story being far into the future, now that I think about it, it also begs the general question: would the uncanny valley phenomenon still hold true with societal and technological advances? Is that sense of eeriness instilled in us biologically or is it just a contextual thing?

Back to my question, and to explain it better: one of my main concerns is how to justify not making these assistants all look like simple orbs or something, instead of the high-fidelity human models I intend to make of them, if their appearance won't affect their performance in any way.

 

[tinkerdan]

On further study of this phenomenon I can't help but think that a larger portion of the problem lie in around what could be a small area where the creation has all the elements that might bring it closest to being human; yet has them out of sync just enough to create this type of feeling or reaction. Since they are expected to evolve so to speak and there is, on the curve, an increasing chance of moving out of that valley it seems the real issue is what to do and how to recognize when they reach the valley of dis-associative human responses and what to do with and for them until they integrate them properly.

This brings to mind Robin William's Bi-centennial Man and the fact that I don't remember having that response to his robot at any particular time; making me wonder if it is just me or that they somehow managed his transformation while steering clear of the valley. Or worse yet if the fact it was Robin Williams created a weird false comfort zone that obviated the possibility.

To add from your further comment:
If it's more an android that has no desire to look human enough to blend into humanity then I can't see it falling so easily into this category. It seems to be mostly relevant when whatever it is whether robot or cgi it begins to look so human that we expect a certain level of synchronization and specialized movement of emotive musculature that we're just as easily disturbed by people suffering from overt affects from Botox.

Sorry but seeing those people in that state sometimes is horrifyingly repulsive.

 

[Ihe]

If it's more an android that has no desire to look human enough to blend into humanity then I can't see it falling so easily into this category.

No no, they are meant to look like us. Their desires aren't part of the equation.

 

[Toby Frost]

On further study of this phenomenon I can't help but think that a larger portion of the problem lie in around what could be a small area where the creation has all the elements that might bring it closest to being human; yet has them out of sync just enough to create this type of feeling or reaction

That's how I see it. The problem (for me) is that you could quite easily avoid this by making the robots look completely inhuman, but giving them advanced minds.

So are the androids visibly human? I can't quite work out where the uncanny quality comes from, if they behave exactly like humans and look like them (unless they stop, break or something unusual like that). Bishop in Aliens is slightly creepy, but he isn't visibly inhuman. He's more like, say, Vincent Prince, in that he seems to be up to something rather than inhuman. On the other hand, if they don't look quite like people, then yes, they'll be sinister as hell (see the Seegson Solutions Working Joe robots.

The society might not find them sinister, but we might (like the geishas in the recent Ghost in the Shell film). That could throw up some interesting possibilities. A story where androids are neither assassins nor slaves would be interesting. Perhaps they could be beautifully made - like artist's mannequins decorated with gold leaf, say, or some kind of very stylish minimal white look - but would still seem unnerving to the reader.

 

[Ihe]

I think uncanny valley concerns human appearances that are slightly off. I don't know if this includes behaviour.

 

[zmunkz]

So, I'm looking for reasonable justifications for this to be so, despite the uncanny valley problem. Any ideas on this?

I don’t think this needs to be a big concern with explicit justifications. If you want this kind of android in your story, do it! In any event, the uncanny valley kicks in when something is visually off (supposedly an instinctual reaction to someone expressing illness or deformity). It occurs for things like the eyes too far apart or too large, something asymmetric or misproportioned, skin tones unrealistic, expressions over or under done, etc.

At the level of advancement you are describing, I am sure these anatomical oddities could be smoothed over. It should be easy to clear the valley either on the non-human side (Wall-E), or the life-like side (Data).

 

[Phyrebrat]

This reminds me of my thoughts when Alien: Isolation came out on PS4; the androids in that look more THX1138 than replica humans, yet Ash in the Alien film is identical to a human in appearance. I think that’s because it’s a plot point in the film; we’re to learn he’s a synthetic along with the crew.

Then there’s the Cylon of 2004’s Battlestar Galactica whose sole journey was to become human so their human appearance was intentional even though they hated their human creators.

pH

 

[Toby Frost]

Wikipedia says that the uncanny valley is motion and appearance, so I don't think it does include behaviour. However, I could imagine that it might: exact repetition of phrases/actions, glitches or freezing up, and so on (this happens in a certain 1970s satire, but I won't give the title as it's a spoiler). I don't know what would happen in a society that was totally used to androids, though. It's possible that they would be considered to be entirely trustworthy, just a bit, er, different, and things we might find disturbing would be written off as "it's a robot thing".

One you're no longer writing about androids as assassins or slaves yearning for freedom (or humanity), some interesting questions crop up. Would something like Westworld be allowed, or would it be seen as too disturbing or injurious to public morals? How do you protect an android (assuming you do) against someone who says "Sleep with me or I'll kill myself"? What happens when Beloved Uncle Robot has done a certain mileage and has to be shut down? All pretty interesting.

 

[Stephen Palmer]

We are hard-wired to intuitively grasp that other human beings are the same as us.
This is a fundamental principle of the evolution of human consciousness, and indeed it is how consciousness works.
We anthropomorphise animals for the same reason. All animist religions use the same principle, except far broader, i.e. rocks, trees etc.
The uncanny valley appears because we simultaneously know that the android is an android, but our conscious minds are struggling with the dilemma of human/almost human. With animals, rocks, trees etc this (very recent) phenomenon doesn't exist.
It is however rooted in exactly the same thing.
So, my advice is: think about the ethical status of your non-humans and relate it to the society they live in.