Uncanny Robots

It’s really remarkable and exciting to learn about the research that’s happening on robots and artificial intelligence these days. It doesn’t get as much press as I feel it should, except when it manifests in odd and surprising places.

Siri, for example. One might argue that Siri — the conversationalist voice which comes with the iPhone 4S — isn’t true artificial intelligence, but frankly, I think it’s a matter of degrees. We humans are difference engines pulling from vast mental databases of information and past experience, making our decisions and our replies to the world. Is Siri any different?

Particularly when we begin taking these artificial intelligences with their responsive abilities and find a way to funnel the vast amount of information, personality, and life on the Internet into them (and find a way to make it USEFUL information. The Internet is many things, but considered as a mind, it’s a crazy person) we’ll very shortly find ourselves with some incredible intelligences.

The biggest mistake I think we’re making, though, is trying to make our robots look human.

I was just watching a video on the remarkable robot replica of Philip K. Dick (who, of course, wrote “Do Androids Dream of Electric Sheep?” which became Blade Runner, a movie full of robots who cannot be distinguished from people, and I have no idea why I’m telling you about it, because you’ve probably seen it whereas I never have. I know that makes me a bad person). Anyway, the replica of Philip K. Dick is remarkable. It has a vast range of facial expressions, adapts to what you’re conversing with it about, and is a pretty fantastic piece of engineering accomplishment.

I also think it’s the wrong direction to go.

The reason, I think, that we want humanoid robots is that ours is an extremely versatile form. We can perform a tremendous range of tasks and skills, we’re very adaptive. It’s why we’re the dominant species on this planet (at least, the dominant life-form that isn’t bacterial). We adapt really well. Therefore, a humanoid robot would not only be able to blend into society, but would be able to preform a huge range of tasks for us in all walks of life. And we would emotionally connect with them and accept them.

Except we wouldn’t. We really wouldn’t.

Trying to make things look Really Human backfires more than it succeeds. the world of CG films has to deal with the fact that they can make staggeringly realistic humans (and, unlike robotics, are not limited by space or the number of muscles and movements they can put in) and yet we still revolt against them, so strongly and consistently that we have a name for it happening: Uncanny Valley. Something becomes just human enough in appearance that we are repulsed by their inhumanity. It’s a problem with eyes, more than anything else, but it’s not just eyes.

(An Aside: I think one remarkable possible use for humanoid robots, combined with drawing a huge database, would be historical robots. The trick is all in the algorithms. If we can find an intelligent enough to sift and funnel the Internet into a cogent body of useful knowledge, imagine if we built a physical replication of Charles Dickens, and then applied that same algorithms not only to all of Dickens’ works, but all of the academic papers about him, the biographies, the letters he wrote and the ones written to him (that correspondence which he didn’t burn, at least). It wouldn’t be perfect, but you could sit and talk to a remarkable impression of Charles Dickens. In this instance, I don’t think the Uncanny Valley effect would bother us. Or at least, I’d be willing to overcome it. Managing this sort of robot would be like realizing Ray Bradbury’s remarkable story about the man who took George Bernard Shaw in robot form into space. It’d be a grand and brilliant experiment.)

So humanoid robots are perhaps a poor idea. At least, replicating humans are. But that said, we do need multi-purpose, adaptable robots. This is the nature of all technology: it converges. The CD player, telephone, address book, desktop computer with Internet access, all converge into the smartphone and the tablet. So we don’t need the individual nanny bot, the Roomba, the car-assembly robot…we need a multi-purpose robot, and one that we can emotionally and socially engage with.

I don’t know about you, but when I think of robots which I am emotionally attached to, it’s robots like R2-D2, and particularly WALL-E, my favorite robot. Or even human-shaped robots like the Iron Giant (although perhaps we would make him a wee bit shorter) or C-3PO. We respond to these robots, we humanize them and love them. If someone marketed a functional WALL-E who could not only perform useful tasks, but who had a responsive database and a personality, it would sell at ludicrous speeds.

We revolt against things which try to fill in their own humanity, but show us something distinctly inhuman, and we’ll pour our own humanity into it and then love it. We do this all the time with some pretty odd animals. We would do it with robots too. I’ve spent many an hour playing with Siri on my wife’s iPhone 4S, and have laughed and delighted in it. Give it a physical form and make it genuinely useful, and people would love it. So long as it wasn’t a human being.

I think when robots do finally enter into the world as autonomic entities, we’ll be surprised to find that many of our movie robot loves will actually be realized in the world. The movies exist, after all, and influence us. Someone will build a robot, and then ten minutes later, someone will have built an R2-D2.

(Of course, he’ll be useless. Watch R2-D2 for a bit. He’s about as actually useful at performing tasks or using stairs as a Dalek, although he is a great deal less likely to EXTERMINAAATE you, so there is that.)

Provided we get the look right, I suspect we’ll all welcome our new Robot Overlords.

And as it happens, I don’t think they’re lord over us. First, I don’t think they’d try. I don’t think the desire to Conquer will necessarily pop up in machines, no matter how much of a database we pour into them. War has so many psychological motivations. If they can generate those, then they are as human as we are, and we have other problems. But the other reason is simply, we could beat robots. Not because of the power of our minds, or because of the power of love, or even because of the power of Keanu Reeves…but go stumble across the Internet for a bit. And then try to figure out how, as a machine, you’d deal with humanity. We are really, really weird.

The Robots are coming, and it’ll be sooner than we think. I can’t wait.


3 thoughts on “Uncanny Robots”

  1. Indeed – a lot of research and effort is poured into mechanical, artificial mimicry of human face for … no real reason. That said, there needs to be some element that “humanity” can be atteched to. As with your example – Wall-E -, features like abstracted eyes / rough ‘face’ should still be a part of these machines, so that they seem less alien and more relatable to humans. In the case of Artoo, well.. I think that his soundbytes are somewhat reminiscent of a living creature’s noises in terms of pitch, direction and responsivity, so that they become the means of direct interaction and personification. A robot with literally no manner of expression (roughly approaching something we as humans can relate to) will never go beyond the idea of ‘a tool, a device’. Perhaps the trick in avoiding the vally is to balance on the left-hand side of it, remaining at/near the peak?

    Then, of course, emains the question of whether we actually want robots to be relatable. Perhaps it would, eventually, be easier for the conscience if they remained featureless?

  2. I don’t think we have a choice but to address the issue of robots being relatable. I briefly mentioned Roomba in the article while making another point, but it applies here too. A Roomba is a dorky looking little disc which hums around on the floor bumping into things like a tiny vacuuming robot drunk. I have a friend who has one in his house.

    And when discussing the Roomba, there has never been a conversation which hasn’t included the comment “and it is so COOL! It’s cute!” or something to that effect. It’s a humming vacuum disc, and we can’t help but attach some humanity to it.

    So unless we’re willing to keep robotic technology on an even simpler plane than Roomba (whichpoint, we’d have the stuff buried in the walls, working without us seeing, which is hardly robotics), I think we have to make it relatable.

    I think what I like about Siri is that if you put the level of responsive personality which Siri already possesses in a humanoid body similar to C-3PO, or WALL-E, we would relate to it without any hesitation whatsoever. You could even make it as workably humanoid as the ASIMO robot. People like that one. *I* like that one, so long as nobody tries to put it in skin (an “Edgar Suit”)

    I think robots are inevitable. All of the proposed functions of robots could, in theory, be buried in the walls and appliances of our homes. But then, all the features of my iPhone could be wired into my house. So why is my apartment not any more sophisticated than an apartment built thirty years ago, but I’ve got this weird piece of science fiction technology in my pocket? I think it speaks to how we like our future-tech, and I think it tells us an awful lot about how we’ll want robots.

    Or at least, I hope so. I WANT a ROBOT. :)

  3. We may all want robots but do we really want to relate to them? I want a robot to do all the crap around my house that I don’t want to do myself. Remember when the upper classes all had servants? Didn’t they train themselves to treat the servants as non-entities and to relate to them as little as possible so they wouldn’t feel bad asking them to do stuff that they themselves felt were demeaning? I can’t see people really wanting to relate to their servants.

    I think what you want is a toy that will fill some sort of need for contact. This toy will make it so you don’t need to make a commitment to another human being, which is messy and comes with all kinds of responsibilities.

    At some point maybe you will want to make a messy commitment to someone. Maybe not. Maybe our robots will alienate us from other real humans. Maybe we won’t miss other humans. Maybe we will finally get our population under control.

Comments are closed.