If you have not read part 1 yet, please click here.
There are two reason why this blog post is so late. The first reason being that I had to wait until I had watched Loopers. The second reason was I’ve been busy with my other hobby. And my final excuse, I read and critiqued one of my brother’s papers. A really great paper, by the way. Look for it in Language.
I’ve already praised the acting, the set design, the camera work, and the special effects on Ex Machina. These are all seamless and well executed. Now, let’s get to the story. The story is straightforward but not simplistic. It has a half-twist at the end but I liked the ending as much as I liked Safety Not Guaranteed’s ending. Let’s talk about the science and engineering.
Passing the Turing test isn’t all that hard. I can write an Eliza program that will pass the Turing test. As a matter of fact, people have already written that Eliza program. Is Eliza sentient? I would have to say no. Forget about self-aware.
I have worked on models of consciousness and cognition with Bernard Baars and others. I’ve worked on models of languages in both theoretical linguistics and psychology. I’ve designed systems of memory for computers that mimic what we think happens in human memory. I’ve even worked at the phonetic level, the very sounds that make up spoken language, in both speech recognition and generation. None of this is or was easy. It can be very daunting. And I’m here to tell you that any one of these areas of research is a lifetime’s endeavor for any of the most brilliant minds on the planet. And even after a lifetime of work, I do not see anyone solving these problems to the level that they were solved in Ex Machina.
We have robots that can walk, though they are noisy. They can see, though imperfectly. They can talk, though they sound tinny. I know people are working on artificial vocal tracts so that the harmonics will match a natural vocal tract but I think that is still years away. The human vocal tract is amazingly complex and resilient. Even when perturbed unexpectedly, humans are still able to make the appropriate phonemes as we talk.
Consciousness and, harder, self-awareness has not come close to being solved. Allen Newell’s SOAR was a great try but as we showed in “A psychologically implausible architecture that is always conscious”, it did not fit the bill of mimicking consciousness forget about being actually conscious. And worse, in Ex Machina, Nathan, the (semi-evil?) (semi-mad?) genius does not want to mimic consciousness, he wants his creation to be self-aware and aware that it has consciousness. In other words, he wants his machine to have personhood.
Let me make a bold statement. I think I could develop a system that is conscious or at least mimics consciousness. It would take awhile, and I would need a lot of funding, and it would only mimic part of what consciousness does. It would not be very complete and I doubt it could pass the Turing Test but it would, I feel, be real. Mind you, the conscious system would be the easy part, and by easy, I mean, really hard, but in order to have a system that goes beyond trivial examples, a language and (abductive and deductive and inductive) reasoning subsystem(s), a memory module, a default mode for 8 hours of sleep complete with dreaming, and a general (abductive and deductive and inductive) learning subsystem(s) would have to be built. This is at a bare minimum. And that would be very hard indeed, and by hard I mean close to impossible.
I get the cold sweats just thinking about the magnitude and complexity of the task.
The brain isn’t some computer you can pluck off the shelf of BrainShack. As a matter of fact, it is a mistake to use a computer to model the brain and it could be a huge mistake to employ computer metaphors to model the mind. The brain is stochastic. The paths through the neurons of the brain are not fixed. The same path, given the same stimulus, is not followed every time. In other words, the brain is very noisy. (Read this paper for more information: The Noisy Brain: Stochastic Dynamics as a Principle of Brain Function) So, I was very charmed that in the movie they used a positronic brain. They didn’t call it that but that’s what they were modeling the artificial brain in the movie after. And rightly so.
I’d like to pause here to talk about the word positronic. A positron is basically an antielectron. That is, a positively charged lepton with the same mass and magnitude of charge as the electron. (The positron has an electric charge of +1 e, a spin of ½, and has the same mass as an electron.) So, for positron, we can substitute antielectron. Thus, positronics is, by analogy with electronics, antielectronics. And this means that it is the same as an antielectronic brain. How using positrons rather than electrons makes the piece of equipment better or the computer better, I don’t know. Anyone care to enlighten me?
(Editors note: If positrons are just electrons traveling backwards in time, particles being just about the only thing in most restricted models of a closed timelike curve that could travel backwards in time, then perhaps a science fiction writer could imagine a temporal computer that did its computing across time. This would make for a very powerful computer. Just thought I’d throw this in for grins.)
We don’t even know what the hardware to mimic the brain would look like. Thus, the science in Ex Machina is basically bogus. And what was that skin made of? Rubber? Some sort of synthetic skin that self anneals? But, in the end, it was a fun romp. Go watch it but don’t think for a minute we are only a billion dollars and twenty prototypes away from true artificial life.
Like Primer, this movie did remind me of my start-up days. Nathan, played by Oscar Isaac, was perfect for the genius founder. The acting was perfect. He drinks too much and just exudes Silicon Valley style. Alicia Vikander, playing Ava, and Sonoya Mizuno, playing Kyoto, were perfect as the androids. Perhaps too perfect to be believable but that was the point of the movie. I want to see a ballerina robot before I believe that the science and engineering is possible in our lifetimes. Domhnall Gleeson, playing the geek, was also perfect.
Now, on to Loopers.
Loopers was a lot of fun. There is no science to speak of and no engineering either. Let’s talk about the paradoxes they use as plots and subplots.
The movie uses the causal paradox to death, also called the bootstrap paradox. For much of the movie, the origins of people and their actions are gotten rid of via this paradox. But this review will have to wait for another day. I’m going to divide this post into two parts. And so ends part one. Part two will be devoted to Loopers.