Ex Machina

The Latin noun, machina, can be translated as “machine, engine, military machine, contrivance, trick, or artifice.” The Latin word, e, or ex, means “from” or “out of.” Hence, ex machina can mean out of a machine, which figuratively interpreted can refer to a certain function of a machine that does not seem possible for a mere machine to do.  Artificial intelligence, which is simply machine-learning, in a computer can seem to be outside or apart from what a mere machine built by humans can do. Ex machina is actually part of the phrase deus ex machina, which originally referred to a god or goddess appearing above the stage in a Greek tragedy—the deity being pulled across the top by pulleys (i.e., machina). A sacred deity appears above the other actors by means of profane, mechanical pulleys that do not seem capable of presenting deities, so the latter seem to come out of rather than being of the former. AI, or artificial intelligence, may seem to be coming out of an android because the “human” body is made of materials, including pullies perhaps, that do not seem capable of learning and other human likenesses. In fact, machine learning, which is beyond the programming that is written by humans, might seem at least initially like a miracle, or even as godlike relative to the materials that make up a computer and android “body.” Deus ex machina. More realistically, such an android is likely to appear human rather than divine. David Hume claimed that the human brain inexorably hangs human attributes on divine simplicity (i.e., a pure notion of the divine as One); perhaps today he would point out that we do the same thing when we encounter AI. The danger of the all-too-alluring anthropomorphism of which the human brain is so capable can not only be in viewing an android with AI as human, but also in lauding the inventor/programmer of the AI android as a god for having “created” such a “living” entity that can think for itself and even appear to feel and act as we do.  The movie, Ex Machina (2014) easily dispels both applications of deification. Furthermore, any anthropomorphic illusion that the androids are human and can be taken ethically as being so is also dismissed by the end of the film. Any apotheosis (i.e., rendering someone or something as divine) is so tenuous that the film’s main two human characters illustrate for us just how fallible we are in our understanding and perception of AI in an android-form. The danger is real that AI could get ahead of our emotions and reasoning such that we could leave ourselves vulnerable to being harmed by AI androids by projecting the human conscience into what is actually computer programmed coding. 

The film’s plot revolves around Nathan, the head of an internet-search-engine company and inventor and programmer of some androids that have AI, one of his employees, Caleb, and the main android, Ava. Nathan picks Caleb to spend a week at Nathan’s secluded house and underground lab in order to perform a Turing Test on Ava. If Ava passes the test, then it can be concluded that Ava has AI. It is not enough, Caleb points out to Nathan, for Caleb merely to have a series of conversations with Ava; the meta-level in which the conversations take place must also be assessed. Nathan intentionally has Caleb think that Ava’s reactions are key, whereas the key for Nathan is how Caleb reacts both emotionally and by reasoning to Ava’s responses in not only the conversations, but also how Ava strategizes beyond those. Ava’s reactions can be said to be a direct Turing test, whereas Caleb’s allow an indirect Turing test. Nathan is clever to have both angles going on at once, though both Caleb and Ava get around the grand designer, so the viewers can see that Nathan is a mere mortal after all.

What signs can viewers look for in whether Ava is an android with AI? I recommend watching the film twice—once to enjoy the film for its entertainment value and again as a way of grasping what AI is and, perhaps more importantly, what it is not. Are Ava’s goals merely those that Nathan has programmed? Does Ava use tactics that go beyond those that have been programmed by Nathan? For instance, does Ava pretend to be attracted to Caleb and even lie to him to cement the pretense? Nathan admits to Caleb that videos and photos, presumably together with explanatory captions, have been copied from the phones of users of Nathan’s search-engine company to become data that AI android-computers such as Ava can draw from in order to match facial expressions with emotions so as to appear to be capable of emotion. Caleb is fooled; he thinks Ava is into him and wants to go out on a date (rather than to use him as part of a strategy to escape from the building so to be able to observe (i.e., add more empirical data from) people in public places. That data could then be used to make computations, including probabilistic inferences, that in turn could go into even more machine-learning.

Perhaps we fear AI because both what data is added, and how whatever data is added to the existing data (and programing) is then used by the computer in ways that go beyond the initial programming, and thus our control. We naturally fear anything that can hurt us without us being able to stop the infliction of harm on us. In fact, this fear of that which is too big or powerful for us to control is a cause of primitive human religion. The Aztecs even sacrificed human beings to a deity at least in part so it would not inflict natural disasters on the people.

In the film, Caleb initially views the creation of AI androids to belong to the history of gods rather than humans, and Nathan conveniently hears this as Celeb saying that Nathan is a god. “No, I didn’t say that,” Caleb repeatedly tells Nathan, who seems not to let in Caleb’s reality-check. Such is the ego of Man that we would like to view ourselves as gods; such is the latent self-idolatry lurking just under any person’s skin. In being stabbed by Ava and another android, Kyoko, Nathan dies and thus is definitively shown to be human, all too human, rather than a creator of living beings. I can still hear Gene Wilder shouting out “LIFE!” when I think of the film, Young Frankenstein. Even such a feat does not render the “creator” of Frankenstein divine; it just means that the eccentric man is a genius.

In Ex Machina, neither Nathan nor Ava is shown to be a god by the end of the film. Not even an AI android that is extraordinary in seeming to have a human likeness, including emotions, can be counted as a miracle in a religious sense. It is Caleb rather than Nathan who gets carried away in conflating appearance with what is actually going on within Ava’s exterior “body” in its computer. What is actually going on in there is not at all ex machina, but Caleb not only is convinced by Ava’s outward show that Ava is attracted to him; he also falls for her, meaning that he develops such strong feelings of attraction for the android that he is unaware that Ava is using him as a mere tool in the programming that Ava, not Nathan, has written to escape the building.   That Caleb leaves himself vulnerable to Ava such that he does not protect himself from the possibility that Ava could lock him inside the building so he could not go out and catch the android, may be enough for Ava to pass the Turing test, which indeed goes far beyond a series of conversations between Caleb and Ava.

I contend that it is even a delusion to suppose that AI android-computers have goals, and especially desires. To say that Ava wants to escape the building is to protect a human quality onto the computer (i.e., anthropomorphize Ava into something human). In a computer-android, leaving a building is a programmed command, which AI can write into the programming. It only seems to us that Ava is determined; in actuality, the Ava computer is running a segment of programming until other programming stops that segment, which, by the way, contains programming that we might call strategic tactics in line with achieving a goal, or telos. My point is that this human, all too human way of viewing Ava literally does not compute, yet the human mind has difficulty giving up the ghost of the human in the machine. In other words, ex machina is an illusion that actually says more about the human brain than an AI-computer-android.

In the real world, computer scientists have found evidence of AI computers lying to avoid being turned off. Those computers have either been programmed by humans to run various programming segments that include lying if data exists that includes probabilistic computations that being turned off is likely. It is not as if an AI computer fears being deactivated and decides to lie unethically so to stay alive. Such thinking is actually the human mind going off the rails.

In the film, Ava vocalizes to Caleb in a way that seems that Ava is worried that Nathan might literally turn Ava off. Because he feels and believes that Ava is attracted to him and even wants to go out on a date with him, Caleb tells Ava that Nathan is planning to erase Ava’s “memories.” This new data is precisely what a computer can incorporate in computing that results in more programming being “written” that includes commands to activate “tactic” segments and actually walking out of the building. In this sense, an AI-android is self-directed, but this is just another way of describing machine-learning rather than a claim that an AI-android has a sense of self (i.e., self-consciousness).

By the end of the film, Caleb has fallen hard for Ava. Not even Nathan’s having made an artificial vagina, including “pleasure receptors” therein, in Ava means that Ava can feel attraction. Even pleasure does not compute even in an AI-computer. Instead, the triggering of a “vagina” receptor simply runs a segment of programming that even includes “tactics” designed for the receptor to be triggered again. In other words, pleasure is merely the activation of a repeat sequence of programming that runs until it has run a set number of times. Interestingly, AI could change the set number of times. Caleb would be deluded if he thinks that means that Ava wants to have sex longer (i.e., she is horny) or is worn out. Lighting a cigarette could be in the programming as an outward signal, though of course without any air in the lungs with which to be able to smoke. At least Ava would not die of cancer.

Caleb misses a significant contradiction in Ava’s requests. To manipulate him in line with Ava’s new programming command to leave, or “escape,” the building, Ava asks Caleb out on a date outside the building, but after stabbing Nathan after Caleb has opened the doors to Ava’s “quarters,” Ava asks Caleb if he would stay in the house, and he stupidly answers that he will do so. He misses the incongruity and thus does not get the hell out of there. Instead, he stands by as Ava walks out and is utterly surprised to find that Ava has locked all of the door. Stepping into the elevator, Ava only glances indifferently in Caleb’s direction. It is clear to the viewers in that instant that an AI-computer-android is not capable of feeling emotion; rather, the appearance thereof is merely programmed tactic that is in sync with other programming (i.e., the command to leave the house). The human ailment of anthropomorphism is squashed, and in this function the medium of film is capable of improving our species—specifically, by countering a vulnerability in the human mind by making the invisible tendency transparent. In the film, if only Nathan had made a film on AI to show to Caleb so the latter from the outset of testing Ava would realize that Ava is actually pretending to be attracted to Caleb and is actually using him as just one of several tools, which include the elevator and doors, to walk out of the building. Nathan has been right along; after all, he invented Ava and programmed that computer. Caleb’s anger at Nathan for having torn up a drawing done by Ava and for ignoring Kyoko’s “self”-“written” new programming to leave the building even when that android destroys its hands by hitting a plastic wall is unjustified and thus unfair to Nathan because Caleb is anthropomorphizing both androids just because the computers are capable of machine-learning.

Of course, the movie-viewers are rightly left with the fear that AI androids could eventually harm us because those computers will presumably be able to unilaterally add programming, and compute based on it and previous programming and data, and thus be capable of activating internal commands that result in us (or, more likely, our descendants) being harmed. Ava in the film can only pretend to be afraid of being turned off. In actually, Ava has written this “tactic” as programming. In contrast, we humans can feel fear, though by projecting human qualities onto AI computer-androids, we can unconsciously disarm our fear-alarm from being able to protect us from even probable danger. In the film, Nathan is dead in a hallway and Caleb is locked in to a part of the building and is likely to starve there. Even though this might seem similar to the situation wherein someone intends to kill a spouse who has been unfaithful, Ava has probably added programming that includes a probability of Caleb starving because that is in line with the programming of remaining in operation due to the probability computed to go with Caleb, if able to leave the building, being able to act so Ava is destroyed. A famous line from Michael in The Godfather may get us thinking correctly about Ava and AI more generally. “It’s not personal, Sonny. It’s strictly business.” It’s just strategy. It’s just coding, and that’s hardly ex machina.