The Machine's Director Takes Us Inside The Worst-Case Scenario For A.I.

The Machine, in select theaters today, is a harrowing low-budget movie about what happens when the military tries to develop its own artificial intelligence. We spoke to writer/director Caradog W. James about his research on real-life military experiments, and the worst thing that could possibly happen with A.I.

The Machine is a really beautiful film, that has a few great strengths — notably among them, its notion that you can't really program artificial intelligence. You have to teach artificial intelligence, by talking to it. This is very reminiscent of some of our favorite A.I. tales, like Ted Chiang's "The Life Cycle of Software Objects." So when we spoke to James on the phone earlier today, we were excited to hear more about this idea.

The thing that interested me most about this film is that this recaptures the idea that you can't reprogram an AI, you have to sort of talk to it and reason it around to your point of view. How did you come up with that notion of talking to AI, and talking to AI instead of reprogramming it?

The inspiration for the project all really came from just my interest in the subject matter, my, kind of, passion for it. I spent almost a year researching the project before I wrote the script. I read every single book I could on AI and robotics and I even read books on quantum computers and quantum theory. And all that research led me to off-the-record meeting with a guy at the Ministry of Defense who's actually building intelligent machines for the government, for the UK government. Weaponized AI. And they're artificially intelligent weapons. Their program uses exact copies of organic brains, so they started off with an exact virtual copy of a mouse brain. So a whole neural net of how information is transferred through neurons in a mouse brain. And they remodeled that and created a virtual world. And when I interviewed the guy he was working on a much bigger project, doing exactly the same on a chimp brain.

The Machine's Director Takes Us Inside The Worst-Case Scenario For A.I.

So it occurred to me after that meeting: 'Well, my God, obviously the next thing they're going to do is have an exact copy of a human brain.' And if you've got the exact copy of a human brain that shares information in the same way, that plans, that, that hopes, that, you know, dreams and wants like a human brain, like the organic original, you know, what is the difference between the original and the copy? Where is the humanity? Is there such a thing as a soul? And that's one of the cornerstones of one of the ideas what I wanted to explore with the film, because, you know, if that's the future of AI, if the future of AI is exact copies of how information is transferred and how we think and how we feel, then, it's not going to be as simple as just telling it what to do, you know. An artificially intelligent brain modeled in that way would have to be taught like a child would have to be taught. And so, you know, obviously you've seen the film, and so that was kind of a keen idea that I was exploring. That is wasn't as simple as just telling a machine what to do. You have to kind of to coerce it and brainwash it into doing what you wanted.

Why was it important to have the first character played by Caity Lotz — the dead scientist whose brain becomes the model for the machines? Why does she need to be a human before she becomes an A.I.?

Well, I mean, again, I think one of the things that kind of frustrates me about bad sci fi is that, you know, it all seems very convenient and their geniuses are kind of geniuses at everything. And again from the research that I did from the people that I met, those pioneers in the field are often very specific. They're very good at one thing, they're not good at everything, and that felt much more real to me, that Vincent would be someone who's perhaps a genius at the hardware, but really what he needed is someone who was great at the software.

The Machine's Director Takes Us Inside The Worst-Case Scenario For A.I.

And that's were Caity Lotz's character came in. She was the kind of, the one who'd had the breakthrough with a machine that was kind of self-learning and self-thinking and self-teaching, and coupling that with his quantum computer, that seemed to me a much more realistic way of of creating the perfect piece of AI, rather than someone who was just able to do everything, you know the kind of clichéd mad scientist who's, you know, brilliant at everything. That just didn't seem to reflect the real world where I was doing research. And so that's why her character's important.

And then the reason that he uses her as, uses her brain as a copy is, obviously, because that's what they were working on, so he's got a head start in that respect. But also because he feels guilty for her death, he feels culpable in her death, and so he uses her face, he keeps her face even though he promises not to because he wants it to upset her. He feels like he deserves to be kind of haunted by her shadow because he feels like he's somehow responsible for her murder.

So, the part that confused me a little bit is that we see when we first meet Ava, she's already built an AI that is close to passing the Turing test, or that pretty much can fool the experts. And yet we still need to scan Ava's brain to create the artificial intelligence in the film. Why isn't the AI that she's already built good enough?

She fails, she fails the Turing test. So her machine could only get so far. I think [Vincent] describes it in the scene as a beautiful piece of programming, but it was still only that — it was something with a huge amount of potential, but it wasn't something that could actually pass the Turing test. And actually one of the scenes that I cut, after a test screening was the scene where the machine passes the Turing test, and the reason that I cut it was because, it kind of becomes self-evident, obviously as the film progresses, that this machine could easily pass the Turing test. But that was what they were working towards together by scanning her brain, by modeling it.

The Machine's Director Takes Us Inside The Worst-Case Scenario For A.I.

One of the problems that [the man at the Ministry of Defense] sort of talked about, in terms of his research and him trying to model a chimp brain, is that there are so many connections, it's so, so difficult to do, such a massive, vast project. It's like a ten-year project at the moment, to map a chimp brain. And really you need something like a quantum computer, that can crunch those numbers in a much faster way, than is available in the moment. So that was the block that Ava kind of met — Ava needed all the resources that Vincent had, and his quantum computer, in order to finish the job that she'd started.

It's sort of interesting the way in which the AI is portrayed as being almost like a child, and yet is also sort of sexualized in this weird way. It's sort of creepy. Is it intended to be creepy? Or is it intended to show two different sides of the same kind of intelligence?

I've noticed a couple of people talking kind of about the sexual elements of the Machine, and it was something that I really wasn't interested in at all. I mean, I guess just having a naked woman I guess as beautiful as Caity, in itself, I suppose, is something that sexualizes scenes. But for myself I never saw their relationship as a sexual one, in any shape or form. I never saw Vincent, you know, looking at her in a sexual way. If anything, he sees her much more as a, kind of like a daughter, rather than someone that he's romantically interested in.

The Machine's Director Takes Us Inside The Worst-Case Scenario For A.I.

I suppose I was interested in the complexity of their relationship. But I was never interested in it becoming a romantic relationship between himself and the Machine. It was something that I really didn't like about the film Splice. It really made me dislike the lead character, in, in Splice when Adrien Brody had a sexual relation with his creation. It immediately made me, kind of, hate that guy. And so that wasn't something that I was looking to do with The Machine at all. But it's interesting how, I guess, some of the imagery is sexy, and I suppose that has kind of, for some people in the audience, that has kind of blurred the lines of their relationship slightly.

Speaking of daughters, so he has his surrogate daughter, who's the AI, and he has his actual daughter. Was there more of that subplot with the daughter's illness and trying to find a cure for her in earlier cuts of the film? That we got to see more of what was going on with her perhaps?

Well, I think there probably was a bit more of her in earlier cuts. But again, I'm kind of happy with where it is at the moment because it was, I was much more interested in his relationship with the machine rather than just kind of wallowing in his kind of despair over his daughter. I mean, part of the research I did for the film is that I met families with severely autistic children and I spent time with them. I interviewed them. And the love of the parents for their children is a huge inspiration for Vincent's character. It's something that was very important to me and very influential in terms of the story I wanted to tell.

The Machine's Director Takes Us Inside The Worst-Case Scenario For A.I.

But it really is just his motivation [for wanting to create artificial life.] There's not much more to explore in that story than his love for his daughter, and his ambiguous feelings about loving her but also at the same time kind of hoping, wishing for a better life for her. And of course the ending of the film is very bittersweet, because he's fixed her but he's also lost her. It would be a question for him in retrospect. Was he happier when he had her love, but she was still disabled? Is it better for her not to be disabled, but for him to kind of lose her as a daughter? And so that interests me about that relationship.

I don't want to talk to much about the ending too much... but this feels like sort of an apocalyptic film, in the end. Is this intended to show the beginnings of the fall of humanity?

Well, I'm very pro-technology. I'm not an anti-technology guy at all. I'm kind of anti, ah, military. And I think that really that's what the film is about is not a fear of technology, so much as a fear of what government and the military will do with these technologies.

I mean, from the people that I spoke to and the research that I've done, it seems quite clear to me that over the next, you know, twenty to fifty years we're going to see very very strong AI, we're going to see remarkable machines that, if they don't do a perfect imitation of life certainly come close. And beyond that, the idea of a machine that truly thinks and feels is something that's very plausible.

The Machine's Director Takes Us Inside The Worst-Case Scenario For A.I.

And so really what the film's about for me is: 'You know, if we're gonna be the parents of these new life forms, we'd better be very careful about the lessons that we teach them. Because if this new species, if this new class of intelligence only has the military to teach them how to behave, then humanity's in very grave trouble.' And so I suppose if anything that's really what the film's about is that sense of, you know, we need to make sure that these advances in technology are in the hands of the right people, not in the hands of people who just want to use it for destructive purposes.

One of the things that was really striking about the film, and really kind of beautiful and weird, was this facility that they're in, this very white, kind of, inhuman looking... with the weird tents almost in the middle of it. Were there things that you did to make that facility seem more scary or more weird? Were there things in set design or the way you lit it or shot it that make it seem more alarming and bizarre?

This is a very low budget production. I mean, it was less than a £1 million to make this film, so a tiny fraction of a usual Hollywood movie. I haven't seen Transcendence, but I think I read somewhere that we're kind of close to, you know, the budget for our whole movie was probably what they spent on catering and transport. And we had to make the whole movie for that. And it was shot in under five weeks, it was shot in kind of four-and-a half-weeks of filming. So we had to make a lot of compromises.

The Machine's Director Takes Us Inside The Worst-Case Scenario For A.I.

But one of the things we were very lucky with is we found this abandoned electronics factory, and that's pretty much where the whole film was set. So, we dressed and augmented this existing location where we pretty much shot the whole film. But that was the only way we were able to get through the schedule because, obviously, if we had had to move to lots of different locations we'd have never been able to get through all the scenes we had to get through just to tell the story.

What gave me confidence of that kind of visual approach is that, in the interviews that I did with the guy who's works for the Ministry of Defense, I asked him, what does your research lab look like? And he explained how most movies kind of get it wrong because, its all chrome and steel and all new, and he said, well the day-to-day reality of working for the Ministry of Defense is that you walk down dusty dark corridors, with wallpaper from the 1970s, furniture from the 1940s and the 1980s, and then there's a one-of-a-kind supercomputer. And that was really what I was interested in was that combination of old and timeless, with very very cutting-edge technology.

Finally, the thing that sort of interests me in this film, which seemed like a weird choice at first, but is really fascinating, is that you have these sort of cyborgs running around, who are sort of, humans who have been augmented. You know, they can't speak like normal humans any more, and they're sort of weirdly inhuman. And then you have the AI who's sort of beautiful and more than human. It seems like there's something in there, there's sort of a message about turning people into weapons, versus turning a weapon into a person, and where the two meet in the middle. Was that a theme that you thought of consciously, the contrast between those two different approaches?

Definitely. And I think that's very perceptive in terms of your describing it in those terms. That was definitely something I was very interested in. I was very lucky that the actress who plays Suri, who's the head of the implant soldiers, the head of the rebellion who sets fire to the traitor at the beginning, she's a famous Iranian actress. And their digital language, I didn't want it to be made up because, when actors try to speak gobbledygook they find it very difficult to have confidence in what they're saying, so I wanted there to be a real structure to their digital language, and so she translated all the lines of dialogue into Farsi, so they're all speaking Farsi.

And that's not a political statement, obviously she knew that I was going to augment that heavily in post to make it sound very very digital, so people wouldn't know. But I wanted there to be a real structure to the language, and for the actors to be able to emotionally invest in the lines, but at the same time I didn't want the audience to be able to lip-read it, and so that was a brilliant addition, that Pooneh Hajimohammadi added to that side of things.

So yes, it's definitely what you're saying, inspired by that idea, and it was also really inspired by the experimental tapes that I saw of people testing mustard gas, British soldiers being forced to inhale, these different types of experimental weapons in the Second World War, and how callous the military establishment is in terms of treating its troops, and it just seemed to me that if they were going to try to push forward this idea of AI, the combination between humans interfacing with machines would also be something they would be very interested in trying to do. And it seems like picking on the most vulnerable, the veterans, the war-damaged, the people whose records say that they're dead, that seemed like a very plausible way that the military would be trying to experiment and push forward their understanding of AI and its interfaces with the brain.

Because I think that's another huge part of where we're gonna to go over the next hundred years, in a hundred years time I think the distinction of human and machine is going to be very fluid. You're seeing it now with Google Glass. Its not a giant leap to imagine something that fits behind your ear that augments your memory or your ability for mathematics, there are already things that emit a magnetic field that'll apparently help your concentration.

If people offered the public something that implanted into the back of your head that allowed you to speak any language on the planed, I'm quite sure people would sign up for that operation, and these are the things which I think are going to be on offer over the next few decades. And once that starts, once people start to augment their bodies in these small ways, then it's really just the start of a much larger augmentation, where we try and push our brains, just like we've pushed the physical side of ourselves, by augmenting it with machinery, I can imagine us trying to push our intelligence and our memory, and our capacity for language, our capacity for mathematics, with machine augmentations. I think the future is going to be very complex in that sense.

The Machine is in select theaters this weekend, and also available on VOD.

Interview transcribed by Madeleine Monson-Rosen.