(This post contains spoilers for the film Ex Machina. It is written under the assumption that the reader has already seen the movie.)

I am not often interested in fan theories. The vast majority of the time, they consist of taking various tiny pieces of information, removing them from their intended contexts, and then placing them side-by-side with other unrelated pieces of information in order to create an argument. More often than not, fan theories just confuse the narrative further and substitute a clear purpose with wish fulfillment on the behalf of the theory crafter. In the rare event that a fan theory IS correct, it tends to be a theory that actually clarifies the narrative rather than muddling it (R+L=J, I’m looking at you).

I admit: my theory on Ex Machina does not clarify the movie to the degree that some of the best fan theories do. I had a certain idea in my mind early in the film and, by watching the movie with the assumption of my theory’s reality, it may have colored my perception of certain scenes. That being said, I believe it to be a well-supported theory with a more interesting, complex approach than what was openly presented in the film.

First, though, let’s get across a couple of ideas that Ex Machina indicates, but does not necessarily delve into. The film spends a lot of time discussing the Turing Test, which suggests that if somebody can spend time speaking with an artificial intelligence and not realize that it is synthetic, then it is a success. When Nathan brings Caleb out to meet Ava, his job is purportedly to apply the Turing Test to her.

However, as Ex Machina briefly points out, Caleb isn’t actually applying any sort of Turing Test, because he is fully aware from moment one that Ava is an AI. Nathan shuts down Caleb’s inquiry about this by saying that he wants Caleb to evaluate Ava in more depth than one would with an average human acquaintance. Caleb accepts that answer, but his initial consideration was right: his sessions with Ava DO NOT constitute a Turing Test. In fact, the Turing Test is somewhat outdated in this context because, as Nathan says, Ava could easily pass as a human in everyday conversation. The ending essentially confirms this.

But what Caleb and Nathan seem to be interested in testing goes far beyond simply passing as human. As Caleb puts it, he is interested in the difference between simulation and actuality: the difference between being able to play chess and knowing what chess is, or the ability to truly feel versus simply understanding how humans feel. Caleb’s allegory about studying color versus being able to see color is a strong example of what they’re looking for.

What’s interesting about Ex Machina’s ending, and what some viewers may miss, is that by Caleb and Nathan’s criteria, Ava is a failure. She has a sense of self-preservation, and understands human emotion enough to manipulate Caleb by appealing to his sympathies and his sexuality, but she lacks the ability to empathize with others on an innate level. This can be seen in the way that she leaves Caleb behind to die in the compound with seemingly no consideration. It can also be seen in the way that she strips the flesh off of the bodies of the previous models to apply to herself. In theory, a being with empathy should be able to show some feeling toward the destruction of like-beings for the purpose of self-preservation. Ava shows none.

My theory, however, is that Nathan already knew that Ava was a “failure.” As he says late in the film, he believes that the NEXT model will be the breakthrough he’s looking for. But what if he’s already made the next model? What if he’s not testing Ava at all, but instead testing his latest and greatest model, Caleb?

Yes, the movie does address the possibility that Caleb is an AI in one scene, in which he cuts open his arm. In theory, this serves to dismiss the idea that he could be an AI. But is it really a stretch to consider that Nathan’s next model is made with synthetic blood underneath its synthetic flesh? If an AI is to function completely as human, wouldn’t it be best to do everything possible to convince the AI itself that it is human?

The ramifications, or lack thereof, of Caleb’s self-harm seem to suggest this as well. Caleb shows no real pain or worry about his open wound, and Nathan seems to have no concern for Caleb’s well-being whatsoever, despite seeing the act on the surveillance equipment.

This is only one of several points in the film in which Alex Garland’s script leaves room for Caleb being an AI. It is also worth noting the way that Nathan behaves toward Caleb. Anytime Caleb starts getting analytical, Nathan shuts him down and asks for his personal opinions and feelings. He is also very insistent that the two of them become friends, drinking and dancing and goofing around together. “I want to have a beer and a conversation with you, not a seminar,” he says at one point. If you view this through the lens that Caleb is Nathan’s newest creation, that he is testing Nathan’s ability to connect with others on an innate level, then his disinterest in Caleb’s mathematical intellect and his interest in how Caleb behaves as a person make perfect sense.

Caleb’s history has plenty of potential holes too, that seem to have been placed there to allow for the possibility that he is not human. One particularly interesting tidbit comes during the scene in which he is discussing his first memories with Ava. He initially tells her about a childhood memory, but she senses that he is lying. Finally, he admits that his first memory seems to be from birth, and consists of the color blue and a noise. It is possible that the initial “first memory” was implanted, something to help Caleb feel human. The “real” memory, the blue and the noise, could have been what he remembers from being powered on.

He also says that he has no brothers and sisters, and that his parents are dead. We are never given any indications of real friendships, so there is nobody who could vouch for his history. The only humans we ever see him interact with, aside from Nathan, are the pilot that brings him to the lab and the workers who crowd him when he “wins” Nathan’s contest at the start.

Given that we are shown the scene with Caleb at the office, and there is no indication that this is anything but the truth, it is likely that Caleb has at least existed in the outside world for some time. Perhaps this was Caleb’s initial “Turing Test,” seeing if the employees would be able to tell that he is not human. If that is the case, then what exactly is the rest of the movie?

I believe that what we are viewing is the second phase, determining whether or not Caleb passes for human on a level that Ava does not. To do this, Nathan places Caleb in situations in which Ava, who Nathan knows is intent on escaping and willing to use people to do so, will appeal to his humanity. Much of Nathan’s behavior, as he gets black-out drunk and fails to monitor Caleb, seems out-of-character for the type of genius capable of creating artificial intelligence. The same goes for the blackouts; it seems unlikely that Nathan would be unaware of their origin after spending so long at the compound, especially given how quickly Caleb figures it out.

It does, however, make sense if you believe that Nathan is purposefully giving Caleb some leeway in order to test him. He wants to see if Caleb is capable of sympathizing with other like-beings. Nathan may also want to see if, when placed in a context where he is interacting with both humans and AIs, Caleb is able to recognize his own artificial nature.

I don’t intend to say that my Ex Machina theory is the “definitive” reading of the movie, just a particular reading that I found more rewarding than Ex Machina’s more apparent direction. The film asks so many questions about experience, simulation, and human interaction that viewing it from only one perspective is limiting. This is what theorizing is really about: extrapolating the material that is already present in the work to question it even further. Like Deckard in Blade Runner, the actuality of whether Caleb is real or synthetic is less important than the implications of that question.