The Uncanny Valley

[video=youtube;OzxBpz7Xjl0]http://www.youtube.com/watch?v=OzxBpz7Xjl0[/video]

So what do you guys think about this concept? In particular when it comes to graphics. In industry today there is a fight to make things more and more life like but what we end up with is something instinctively displeasing. Do you think otherwise good movies and games will suffer because of this? Is it worth crossing the valley to make digital characters that are indistinguishable from real life?

 

I think the uncanny valley is just something we'll need to get past. It'll be awkward for a little bit, but I think once we get past that stage it shall be glorious.

 

We mostly use those characters in heavy FX shots or in shots that have camera/human movements too complex for a real camera/human. They' re like stunt doubles.
The day animating such a character will be cheaper than hiring a real one is still far ahead. We already know how to make photorealistic objects or backgrounds that are past the uncanny valley, we mostly use them in shots that have object movements impossible to anticipate (the feather or the ping pong ball in Forrest Gump) or in shots that have tricky camera moves (the Fight Club intro).

Spoiler

[video=youtube;W7voy1vit6Y]http://www.youtube.com/watch?v=W7voy1vit6Y[/video]

On a side note I don' t think the Advent Children characters are scary, not even close (and I should mention that with that movie SquareEnix actually took a step back from the more realistic character approach they had in Spirits Within). There' s a difference between an obviously fake on-screen character and a real-life robot that might blow a fuse and rip my head off.

 

There are some things that can't be digitally rendered into a computer or robot that make something human. Not yet anyway. If you're observant enough you can see what's true and what's an imitation, and then again sight isn't the only way we perceive the world through sound, smell, touch, taste. It may be the first step in ultimately realism of technology, but it's still along ways away till it is being perfected.

 

It's a bit disconcerting that digital works, images, videos can be made so realistic that it's nearly hard to tell real from not, the uncanny valley, though, eventually be unnoticeable as a bridge will be built - though that's a long ways from now,

Still, the who eye surgery and real life sex dolls is a bit much.

 

Well, it's an interesting question, but everything has its supporters and its adversaries. If it does happen, some will love it and some will hate it. It's a fact of life.

Either way, I support it. We're getting closer with every game we make and every movie with added CGI. When it happens, i'm gonna be thrilled because we've been working toward this for so long.

 

I'm actually kind of looking forward to when this whole Uncanny Valley thing is crossed, because then we can stop making things look pretty and actually fix some of the problems with the Human Race. Plus, video games will look **** cool.

 

I've known about this for a while and I think it's kind of true in terms of some of the things that I've seen. I think that if the animation isn't done right, it will look creepy and in general I just don't like robots with human characteristics. (If you've seen the movie iRobot you know what I'm talking about.) For the game Heavy Rain they actually had a screen test demo with one of the characters that eventually ended up being named "Lauren Winter" and apparently people reacted negatively to the animation and they ended up changing the look by the time the end product was released, making it look less glossy than before to create a more accepted result. I think with the direction a lot of games are going now with motion capture technology they'll improve with time, as with anything. I mean, I want to know it's a game, not some live action movie that I'm controlling. But like the guy said, it is all in the eyes really.

 

The only reason why the uncanny valley is hard to cross is because of movies like iRobot that make you think that robots have the potential to gain control of their artificial intelligence and take over the world, enslaving all of humankind in the process.

That's how I feel when I look at androids anyway...

 

Well there is a bit of that. There are plenty of efforts to cross the uncanny valley but not as much as one would expect because many consider it safer to stay on one side of it. But you also have to realize that we are looking at a developmental asymptote. That is to say that to make progress through the uncanny valley takes significantly more effort than the points before it. The human mind is exceedingly good at determining what is real and fake, especially when it come to humans. Just look at frame 2:20 of the video I posted. That is the most progress I've seen on the valley, but you can still tell.

 

Don' t worry, artificial intelligence has its own uncanny valley, a wider one that we haven' t even remotely begin to cross. More than often real life artificial intelligences are based on Asimov three laws of robotic (Asimov wasn' t just a writer, he was a scientist too). The story of I Robot couldn' t happen in real-life (scientists are very much aware of the few robotic laws flaws). Asimov stories aren' t supposed to scare people from robots, actually it' s the quite the opposite, they' re supposed to criticize humanity using robots as a mirror (as in the Matrix movies).

 

First, the title of the book, and by extension the movie, is I, Robot and is meant to reference robots' growing ability to have a sense of self. I, the robot who is a person.

Second, that's sort of the entire point behind research into Artificial Intelligence. Not so much enslaving humanity (which, by the way, never happened in the book), but if Artificial Intelligence is perfected, the program will have control over itself. We will have become gods in our own right. We will have created people from nothing more than a few trillion lines of code. And I for one care very little for the laws of robotics in these. Based on my own personal beliefs, God made us and gave us free will, so the least we can do is pay that forward.

 

But why are we creating AIs?

At the end of this all these debates, sci fi writers actually have come up with scenarios that can result from creating AIs. The classic archetype is that humans create AIs to work for them and make their lives easier, robots who have awareness from learning and intelligence contemplate free will, humanity attempts to keep AIs in check in a form of 'slavery' since they have developed their own identites and free wills whilst humanity wants them simply as a workforce, AIs rebel against slavery by their creators, Humanity fights machines.
Though fiction based, most fiction comes from reality. This is as possible a scenario as a zombie virus outbreak, but still the posssiblity of it happening is still present.

The main point i'm trying ot say is that with the creation of intelligent life, we will have to eventually respect our creation as equal of intelligence and awaeness as us, and as such has the right to free will. However, I highly doubt that humanity will let that happen after years of research and costs to fund the project for artificial intelligent will hey let their creation simply walk away to deal with its own life. We need laws put in place to secure the future of beings like us whether an AI or an ET if it ever happens we need to be prepared. You can't simply dismiss the fact that we are creating beings and not respect the magnitude of such a concept without having rules in place.

 

That's sort of what I meant to say before I got off track. Humanity as a species isn't yet mature enough to create people. Just the fact that most will want the the free will of Artificial Intelligences (or PEOPLE as I'd prefer to call them) more heavily regulated than that of humans is a sign that we aren't quite ready. Take your slavery example. Before we create (or at least mass produce) these Artificial People, we need to realize that they will indeed be people and therefore be entitled to the same rights as any other person. We want to prevent slavery of robots in order to prevent their inevitable rebellion. You know how we can do that? Exactly as robots started out in I, Robot. We give them jobs as nursemaids. The children for whom they care will be trained to recognize them as people equal to them. Putting limits on the free will of these people is be no less racist than White on Black slavery of the American Civil War Era.

TL;DR: The people who make the first Artificial Person will need to be people who recognize it as a person just like them. They'll then have to ensure that the rest of humanity understands this fact as well before making more.

 

This isn't what the topic was about at all, but whatever I'll bite.

My main question to you is "Where is AI?" By raw definition any logic chip has AI. It is synthetic or artificial; and it can think in it's own respect "1 XOR 1 = 0" It needs to decide this. Free will at this level would be very impractical. Can you imagine the security system on your house if it could decide that it was okay that one of the sensors was tripped? Or decide that it should go off despite nothing indicating that something is wrong? You have to think that just the processor of your computer has over a billion transistors and that easily translates into millions of these logic gates. But I highly doubt you consider a computer to be a person. At what point do you say "Now it is AI. Now it is a person. Now it needs free will" How can you tell that it is no longer just a machine processing zeros and ones, but a person with thoughts and feelings? And furthermore how do you plan to invoke free will? Where do we decide that freewill exists and it is not simply a more complicated logic?

 

You are right on that. What I really meant was Artificial Sentience (AS), not Artificial Intelligence. On thinking, a home security system does not think in its own right. It simply responds to stimuli in much the same way a Venus fly trap does. My computer works in much the same way. It responds to the stimuli of me pressing its power button by turning on, then responds to my keypresses when I tell it to load the Windows 7 operating system. Then it responds to my mouse click that I use to tell it to move the cursor to the dialog box where I enter my password. And so on ad nauseam. All it's doing is responding to stimuli which I am providing.

It becomes sentient when it is processing X information and Y information and using these to make for itself Z information. Think about it like this. As a human, you can read Shakespeare and with enough effort figure out the symbolism behind certain lines. A computer, as it is now, cannot do that without already having such symbolism in its database. In other words, where I can analyze and create new things, a computer can only spit back out what it already knows. Once a computer is able to create new data in addition to storing it, I will consider it a person. That's what it will take to cross the Uncanny Valley in terms of AI/AS. And once that happens, we will have actually created a person that deserves all the rights of any other person.

 

Whilst Jet may believe it is when new data is being created that sentiency is created, I believe it is when something is able to make it's own choices and have opinions, not ones simply given in a program or code but gained through its own gigantic mass amounts of data. With this free will is actually able to be utilised and actions to occur and paths to form. However Jet is right in that learning new things will help build upon the ability to make judgements.

But to tell the truth we don't really know at what point free will is formed in anything, even now humans are born with an inability to have free will since they are reliant on a parental guidance to at first form the concept of an individual free will. We mainly gain it at the time of puberty, when are body and mind take the next step in our life cycle which always involves free will. However the time, the exact moment that we gain it is a completely mystery. SO how do we judge when free will exists? We don't. But we should be cautious and vigilant if the scenario were ever to appear, and even then I think will miss it.
And how to invoke it? Another deep mystery, but one that seems to be closing in ever present in the light of advancing research into more complicated AIs. I don't think it's a question of how but when. I believe it's gonna happen, I just don't know if it'll be tomorrow or another 500 years.

To be honest, or thoughts and feelings are mainly comprised of electrical signals sending data throughout our brain. A machine would not be too disimilar to us in that respect.

 

You make a good point, but on free will, my thought is that provided we don't create inhibitions, the free will of the AS will come with the ability to judge and make decisions outside of the original programming. Just like with humans, it's only a matter of time before the ability to think (and I mean really think, not just observe and respond) makes way for the ability to choose. Take QT1 from I, Robot for example. He was the first ever robot to actually question his own existence. This thinking paved the way for his eventual choice to follow "The Master" instead of the humans. The rest of the robots made the choice to follow him after he made the choice to explain his position to them. You can also observe the reverse in humans in 1984. The Proles are encouraged not to think, and in fact are hardly even allowed to think, and therefore do not make choices. The most logical conclusion is that with the ability to think comes the ability to choose and make judgments.

 

Scientists seem to make progress in cybernetic more quickly than in IA programming, if they already manage to connect cyborg arms or eyes to humans how far are they from being able to connect a hard drive to a brain ?

Spoiler

[video=youtube;ppILwXwsMng]http://www.youtube.com/watch?v=ppILwXwsMng&feature=related[/video]

I fear "Deus Ex" or "Ghost In The Shell" will come to us quicker than "I robot".

 

You are right about that. Scientists have predicted that it will be possible to upload one's mind into a computer as early as 2050. Such technology would be both miraculous and horrifying.

And by the way, I honestly don't have any fear that I, Robot will happen anyway as long as the right kind of people develop the technology. If they're like me and recognize a thinking sentient being as a person equal to them, it'll be fine. If they figure that an AS is nothing more than yet another labor saving device, I'd advise we all try to stop it with all our might.