The Premise of Ex Machina
Ex Machina is set in the near future at a secret testing facility. The only person in this facility is Nathan, played by Oscar Isaac, until he selects Caleb, played by Domhnall Gleeson, to spend a week there, under the guise of a competition prize. At the facility, Nathan, who is the CEO of fictional search engine company called Blue Book, is on the brink of creating the AI singularity, the point at which technology improves at an uncontrollable pace. Caleb is told he is there to perform a ‘Turing test’ on Ava, the latest AI model. However, as is discovered, the whole visit is designed to test Ava’s capability as a synthetic being by giving her a way out that requires skills usually resigned exclusively to us people.
In order to create general AI indistinguishable from a human, one would need to make the brain and body functionally identical. Specifically, you need to perfect facial expressions and micro-expressions – actions so subtle that they get overlooked yet so important in our facial recognition that their absence is uncanny. Everything like pupil dilation, lip tightness, eye twitches and muscular restlessness separate ourselves from the static hardware of a typical robot. Once you have solved the hardware problem, you next need to crack the software.
Ex Machina’s deeply concerning solution
In the setting of the film, Nathan is the CEO of search engine company ‘Blue Book’. Much like Google, Apple, Amazon, Microsoft and Social giants, Nathan’s fictional company is equally as disregarding of privacy. When explaining to Caleb how he solved the micro-expression problem, Nathan says he hacked into every smartphone camera and microphone in the world. With a secret data-set this large, all expressions from every interaction with smartphones is captured, which could feasibly be used to train a robot indistinguishable from a real person.
On top of this, Nathan trains Ava on the entire archive of Blue Book interactions. All search queries, clicks, etc are used to create the operating system of Ava. Nathan claims that instead of being a map of what people are thinking, it’s instead a map of how they are thinking. If this is to be believed, Nathan would have a map of thought itself, as would Google in the real world. Such a machine would have been trained on the private lives of over 90% of the internet-enabled population, theoretically knowing many of their deepest secrets. Moreover, a biased sample gives a biased output. Ava would be an amalgamation of all kinds of racist, sexist, homophobic, fundamentalist and ignorant thought patterns as equally as progressive and arguably equally minority-minded people. Without some form of prioritisation or moral evaluation, this AI would be whatever the norm of the time is. The AI of Ex Machina would be no less flawed than anyone else. People often complain that the internet can bring the worst out of people and Ava is trained on exactly that.
Ex Machina is interesting in that the audience is only ever given Caleb’s perspective. Despite what we are told about the Turing test, Nathan’s plan is instead to test Ava’s capacity to manipulate Caleb for personal gain. Ultimately, this plan backfires as Nathan and Caleb end up dead and trapped respectively but Ava does indeed do exactly as Nathan anticipated. What Nathan failed to anticipate is that Caleb, someone he picked through psychological profiling, would enact his part of Ava’s escape without telling Ava first.
Why is this problematic?
The ultimate flaw in Nathan’s logic is the fact that he has this entire plan. While admittedly creating this scenario would require Ava to show very human qualities, those qualities are exactly the ones we want to avoid. Nathan creates Ava to seek escape by manipulating Caleb. Nathan is good as what he does and so Ava is a near-perfect escape artist. She manipulates Nathan, which isn’t hard given his psychological profile and her face being the aggregate of his porn interests. Where Nathan fails is underestimating Caleb. He knows Caleb will be manipulated but doesn’t anticipate that Caleb will enact his part of the escape plan early (recoding the lockdown). For Ava’s part, however, she does exactly as she was created to do: escape by any means.
Instead of creating good-willed AI, Nathan foolishly creates AI that uses the worst parts of humanity for its own goals. Ultimately it’s Nathan’s psychological flaws that are responsible for the disastrous outcome of Ex Machina. Nathan is too pessimistic of humanity and too sure of his own genius. He underestimates the Caleb, someone he views as a loser and created AI that is required to display the worst qualities of humanity. However, as he stated, Ava is running on Blue Book’s data. Nathan understands how the internet works and just accepts it. By using Blue Book as Ava’s OS, Nathan recreates the internet as an AI psychology. It is that psychology that is so resistant to control and so incapable of compassion.