Terminator, Turing and AI
So I thought the second to last episode of the Sarah Connor Chronicles was really excellent, the scene were she walks out of the apartment with the two Russians not only was some of the best TV I’ve seen recently but it retroactively made the series great by contextualizing Cameron’s attempts to “humanize” herself. (Compare and contrast with Star Wars Eps I-III which retroactively make the eps IV-VI worse by, among many other things, turning the force from a mystical energy to some bacteria in people that you measure with the Ghost Busters’ paranormal activity-ometer.) I spent some time wondering why I found that scene so particularly interesting… here are the sweet, sweet fruits of that time. Thar be spoilers ahead, be ye afraid of such things, turn back now. Go watch a hamster eating some stuff.
So the scene in question, if you don’t remember, is when Cameron (Summer Glau) goes undercover at a ballet studio. There’s much talk about dancing and soul and much reflecting about being human. Long story short, she’s a good dancer and wins the trust of the instructor to bring her to the guy she’s looking for who is in hiding and being pursued by nameless bad guys. Using her growing human empathy she convinces the dude that she can help him once he gives her the 411 that she’s looking for, the process is hurried by the fact that nameless baddies have pulled up outside the building and are coming in, we’ve (including the instructor) already seen Cameron pummel one such baddy. The guy gives Cameron what she wants who then without a glance back or even an expression on her face, turns and walks out the door… walks right past the bad guys running up the stairs who get to the instructor and the guy and, presumably, put the kibosh on them. That was so awesome.
Typically a story of this nature has the robot in question slowly becoming more human and this typically means more lovable and more trustworthy. Someone you can like, understand and relate to. It seems to me that Terminator is very notably not going in that direction. That Cameron’s attempts to become more human simply are a mechanical understanding of what it means to act more human without actually understanding the underlying emotions - or maybe not caring about those emotions. That despite the fact that she grows better at fooling humans into thinking that she is one of them, in fact she isn’t human at all and remains completely machine. It is the purpose of all those advanced terminators - to be able to infiltrate the good guys - they need to be able to seem like one of them.
I think that’s a pretty neat premise. It puts the Turing Test in an interesting light - the test for artificial intelligence summed up from wikipedia “a human judge engages in a natural language conversation with one human and one machine, each of which try to appear human; if the judge cannot reliably tell which is which, then the machine is said to pass the test.” Or more generally that the indistinguishable appearance of something is just as good as being that something. That a machine that can simulate human emotion is just as good as having that emotion - but is it? Obviously, the Turing Test wasn’t meant to cover that - but it seems a reasonable extension and an interesting thing to think about.
It’s like when I read that Kurzweil book, The Age of Spiritual Machines. And while I think that the man is an ass who condescends to you in his writing and I no longer agree specifically with the idea that the human brain is a completely predictable and replicable machine, the general notion that at some point computers get fast enough to become intelligent which brings us to a point where we can no longer imagine what will happen seems very plausible. Oh, I also can’t wait to be a nano-swarm. But getting back to this.
A lot of people believe that when machines achieve this intelligence that somehow it will be similar to human’s. That it will be modeled on a human brain, perhaps. I wonder if that is a human-centric view of intelligence - that is, why would it necessarily even be recognized as intelligence by us? If it’s complexity grows to the point where we no longer understand what it’s doing (this is happening, hey man look at Vista! Redmond has no idea what’s going on there… ;) - if things happen who’s to say that’s not a sign of intelligence. Maybe the machine learns to act like a human, but if it is so alien to the emotions why would it care? Why would it even be pre-occupied with anything we did? Without an evolutionary impetus to survive and procreate, what will its driving force be? Why be Skynet, then, and not simply contemplate this very interesting stream of 1’s and 0’s? Y’know?
So getting back to SCC, I love this direction they’ve taken with Summer. I think playing around with people really falling for her and her growing understanding of humanity while showing these clues that she doesn’t just have a cute and quirky misunderstanding of humanity but rather is in fact monstrously inhuman is really great. I’m just saying. Terminator’s cool.
