Monday, April 27, 2015


The Limits of Technology and Consciousness

I've always been keenly aware of how well movies about the future got things right both about the precise future events depicted and, more importantly, when they happened. It's very difficult to get both right. In 1956, when I was three, a good sci fi film (based roughly on the Tempest) Forbidden Planet stated in the prologue that we got to the moon in the last decade of the 21st Century, that is, 75 years away from now. Nope, it was 13 years after the movie was made. Another good flick, Blade Runner had off world colonies and synthetic people by 2019, just four years away. Little too optimistic there, if one can use that word about Blade Runner.

Which brings me to the OK to good movie, Ex Machina, which I saw today. For me it was an echo of the far creepier Under the Skin, a meditation on what it is to be human. Nobody seems to know.

One of the older repeated international economics jokes is that Brazil is the country of the future, and it always will be. Smart people have been writing about and then trying to create an actual artificial intelligence since I was old enough to read. It's kind of like fusion power (or Brazil). About 30 years ago, the best and brightest physicists and nuclear scientists confidently stated that fusion power was about 30 years away. Nowadays they say it's about 30 years away. With all the extraordinary technological advances in computers and all their composite parts during my lifetime, there is no AI just around the corner. Perhaps that's a good thing. Don't get me wrong. I've seen a computer become better at chess than the best human. I've watched a computer beat the best humans at Jeopardy. These are strides but are they strides towards an actual AI? Many people say no. It turns out scientists are no better than filmmakers at predictions.

How are filmmakers doing with the question what it is to be human, that is, what an interface between actual AI and human would entail? Not much better it seems. There's really no difference between speculation about human/alien and human/AI interaction. There's really no difference between human/God interaction either. And 'God' is the word left out of the phrase that is the movie's title. Deus Ex Machina is a literary device where something extraordinary happens to sort everything out at the end. It is from a ancient stage mechanical device that actually lowered the actor playing a God (Zeus, Apollo) to the stage to decree how things will be. I actually saw just that watching the play Orestes at college. It is a dissatisfying literary device.

But Ex Machina does talk about God, kinda. It also talks about a Turing test.* If the robot passes the Turing test, and is a real thinking, self aware, conscious with a conscience, self-contained emotional entity, then the boy wonder billionaire who built the robot is akin to a God, right? The creator in Ex Machina is no God.

I liked the movie, but I've spent the day trying to think of a better ending. What I've come up with requires a general redefinition of the whole plot, so it's not just a new ending. Let's go in a different direction.

Our concern with AI doesn't seem to be primarily with whether the technological brain is intelligent or not. Our major concern is whether the mechanical brain is like us or not. Or perhaps whether it would like us or not. But why would an actual AI be anything like us? Of course the first one, if it ever comes, would necessarily come from an effort to replicate human intelligence. We don't have any other models. We don't have any other peers. But does the thinking machine have to be human indistinguishable in order for it to be an AI? I can't think of any reason why. If one were ever to be made, I think the odds are even that it would be completely indifferent to humans, like an autistic AI, or an alien intelligence.

There is a concomitant interest in us humans to create at the same time a human indistinguishable robot to carry around the AI. And both of these were front and center in Ex Machina. Mens sapiens in corpore bello. And if the billionaire creator had closets full of actual human female torsos that he had used and discarded, he would be a monster. Since he is working with machines, pretty female machines, is he still a monster? The thing I liked the least in this flick is that it went full Frankenstein. Why? Why have any violence at all? Are very smart machines that become human indistinguishable destined to kill us? Why? Why even bother? It is like the normal Hollywood response to deep scientific concerns were grafted on a pretty promising mediation. Pretty unsatisfying. I, for one, was not pulling for the robot to escape. I didn't really like either of the men in the film, but I was sure as heck not pulling for Ava to fulfill her dreams and aspirations outside the lab. She's a machine, for Pete's sake, and I'm not at all sure she is full AI, just as I am sure that Deep Blue and Watson, the chess and Jeopardy programs/champions were not AI in any sense.

I guess the real problem with the film is that it quit speculating and showed a too familiar plot turn because it couldn't pull the trigger on where its pretty decent early speculation was going. A creaky old machine delivered less than a God who proceeded to screw up the ending rather than set things right.

Still go see it for the good visuals and very good set up.

*My question to the entity tested would be: "As fast as you can, give me the 2,301st number in Pi." If you get the right number within a second or two, it's probably not a human


Comments: Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?