Brett Gregory interviews Dr Paula Murphy (Dublin City University) about her new book, ‘AI in the Movies’
Hi, my name is Brett Gregory, and this is a podcast interview for the UK arts, culture and politics website, Culture Matters. What follows is an extremely interesting discussion with the author of a new book called, ‘AI in the Movies’, which has been published by Edinburgh University Press.
Brett: Hello, and welcome. Please, introduce yourself.
Paula: Hello, my name is Dr. Paula Murphy. I lecture in the School of English in Dublin City University, Ireland, and I specialise in Modern Irish Literature and Film, and popular film, especially film representations of artificial intelligence, which is the topic of my book ‘AI in the Movies’.
Brett: And what specifically inspired you to begin writing this particular book, Paula?
Paula: I first started thinking about AI, and how it is represented in film, when I saw the film ‘Her’ in 2013, and it fascinated me. It’s a film directed by Spike Jonze, starring Joaquin Phoenix as Theodore Twombly. It’s about a lonely man who has separated from his wife, and finds love with an artificially intelligent operating system called Samantha, voiced by Scarlett Johansson. He uploads the operating system on his phone, and when Samantha begins to communicate, he is startled by how human she is. She’s clever and witty, she’s supportive and encouraging. And as soon as she comes into being, she begins to change and develop, to have aspirations and longings, an emotional and sexual life. She goes on a journey of coming to terms with herself, her abilities and her limitations, that leads to her finally leaving Theodore. In fact, ultimately, leaving the human, material world, entirely.
Brett: Yeah, I’ve seen it. Very involving, resonant. A key film for understanding the first quarter of the 21st century, I’d argue.
Paula: At one level, it’s a very human story. If I described the trajectory of that romance to you without mentioning that Samantha was an AI, it sounds entirely plausible. But, at another level, there are ominous notes in the film about human relationships with AI. Theodore thinks of her as human: it is her humanness that he falls in love with, but he understands at the end of the film, that she was only showing him a part of herself. He tries and fails to keep up with her intellectually. The speed at which she processes knowledge is far beyond his capability, even his understanding. Human thought and communication are soon frustratingly slow to her, and she excuses herself at one point in the film to communicate ‘post-verbally’ to a dead philosopher whose brain she and other AIs have artificially reconstructed.
Theodore’s humanising of her means that he is genuinely bereft when he discovers that she is having relationships with multiple humans and AIs, many of whom she is romantically involved with. He so easily adapts to her lack of physical presence – her lack of a body – as do his friends. That casting aside of the material world, which is the world that we as humans are irrevocably tethered to, struck me as dangerous. And he happily ignores the access he has given her, or not given her, to his personal data, such that, without his knowledge, she puts together a book comprised of letters he has written and sends it to a publisher, posing as him. So there were lots of aspects of the film ‘Her’ that got me thinking, and it motivated me to start looking at other films where AI had been represented and, eventually, to try and watch and analyse them all, and try to trace the recurring themes and patterns, narrative and visual, across the decades.
Brett: Excellent. Right, now, many people, ill-informed by the mainstream media as usual, generally perceive AI to be this single definitive category which encompasses absolutely everything computer-related and/or computer-generated. But this isn’t the case, is it? For example, in your book you introduce us to ‘affective AI’.
Paula: Affective AI can identify human emotion through, for example, facial expression, gestures, or voice intonation. In the real world, affective AI is used by companies in things like market research, customer service, and the automotive industry, to gauge customers’ emotional reactions. Unlike the real world, in the films analysed in this book, the AIs are usually self-aware, and emotionally complex, capable of not only identifying human emotion, but reciprocating it.
Brett: And what about ‘ambient intelligence’?
Paula: Ambient intelligence is AI that lives in our environment with us; it is there is the background in the form of a smartwatch, a digital assistant, or a robotic vacuum cleaner. Films imagine this type of AI too becoming self-aware and autonomous, like the smart home assistant ‘Tau’ in the Netflix film of the same name.
Brett: I haven’t watched ‘Tau’, but it’s now on the list.
Paula: Humanoid AI robots can be robots that are shaped like humans in the sense that they have a torso, a head, arms and legs. In film these range from utilitarian police droids, like ‘Chappie’, to robots that are indistinguishable from humans, like Rachael in ‘Blade Runner’, for example. In the real world humanoid robots take in a similarly broad range of types, from Boston Dynamic’s Atlas robot to the robots produced by Hanson robotics with an uncanny similarity to humans.
Brett: And there are ‘digital AIs’ as well?
Paula: Digital AIs are AIs that do not have a robot body of any kind. In films they can be housed in a computer, like Edgar in the 1980s movie ‘Electric Dreams’, or on a spaceship, like HAL 9000 in ‘2001: A Space Odyssey’, or they can exist online, like the Puppet Master in ‘Ghost in the Shell’. We are familiar with digital AIs in the real world too: the voices that speak to us in customer service chatbots, or digital assistants. The important difference between real-world AIs and the ones discussed in the book, is that all the AIs in the book are film representations of strong or human level AI: they are autonomous individuals with their own sense of self, their own desires, ambitions and moral code, and we don’t have that in the real world, not yet anyway.
Brett: Erm, fingers crossed. And what about ‘hybridity’?
Paula: Hybridity is a really interesting feature of artificial intelligence representations in film. This describes entities that are made up part human, part AI components. I don’t think there are real world comparisons to the type of hybridisation we see in film. For example, in ‘Terminator: Salvation’, the character of Marcus is a hybrid figure. He is a human who is selected by the artificial intelligence Skynet for ‘modification’, and is given a cybernetic heart and a machine brain that syncs with Skynet. For me, these hybridised characters are among the most interesting in artificial intelligence film, because they illustrate how complex and entangled the relationship between humans and AI can be.
Brett: Now, couldn’t it be argued that the use of AI technology as a storyline, character or trope is just another Hollywood show business tool used to draw in and spook the audience. For example, aren’t sci-fi thrillers such as ‘Westworld’ from 1973 or ‘Demon Seed’ from 1977 just simply B-Movie ‘creature features’ like ‘Frankenstein’ from 1931 or ‘The Thing from Another World’ from 1951?
Paula: It is absolutely true that film has always harnessed technological innovation to bring its viewers films that are more realistic or entertaining or exciting. In the films analysed in this book, artificial intelligence hasn’t been particularly evident as a technological innovation, like sound, or CGI. But it is there as a trope, in storylines and in characters, and certainly, in the main, it is there as something to be afraid of, something that we don’t fully understand, that is potentially more powerful than us, and which frightens us.
In this sense, many AI films, particularly the older ones from the 50s, 60s and 70s that you’ve mentioned, have a lot in common with B-Movie ‘creature features’. The AIs, like the monsters, are presented as aberrations, and the characteristics that they share with humans makes them more terrifying, not less terrifying, dredging up the horror of the uncanny. Most AI films are anthropocentric: they put the human at the centre. Because of this, the AI often functions as a mirror to distasteful human attributes and emotions: ambition, jealousy, revenge. Just like the B-Movie monsters, AI can represent those parts of human nature that we might wish to remain hidden. AIs share another characteristic with B-Movie monsters, and it is an interest in where the dividing line is between ‘them and us’. A key question that is asked about ‘The Thing from Another World’ is: ‘is it human or inhuman?’ That same question is asked about AIs over and over again in the history of AI film.
Brett: Of course, it could be argued also that using AI to interfere with the physical condition of human beings, or even raising them from the dead, is unnerving, unnatural and unholy, like necromancy or zombification. For instance, movies like ‘The Six Million Dollar Man’ from 1973, ‘Robocop’ from 1987 and ‘Upgrade’ from 2018 portray a semi-posthumous protagonist enhanced by an exoskeleton, and they are healthy, empowered, death-defying, and immortal as a consequence.
Paula: Well, ‘The Six Million Dollar Man’ and ‘Robocop’ wouldn’t fall into the category of an AI movie in terms of the parameters of this study. Those characters are more cyborgs than artificial intelligences.
Brett: I see.
Paula: In terms of ‘Upgrade’, yes, there is an AI there called STEM who inhabits the body of the paralysed man. ‘Upgrade’ is similar to films like ‘Transcendence’ and ‘Chappie’ in which a strong AI is used to extend or augment human life. But you’re perfectly correct about all of the films that you mention presenting technology as a panacea to ill-health, injury and even mortality, and there certainly is something deeply unsettling about that idea.
Brett: Excellent. I’m generally on the right lines then.
Paula: On the flip side of that, there are AI films that present AI immortality as a problematic obstacle to the humanness that the AIs desire. A great example of that is ‘Bicentennial Man’, starring Robin Williams. At the beginning of the film, Robin Williams’ character, Andrew, is a robot, but by the end, he has aesthetically and biologically transformed into a human being. His immortality is the final obstacle to him being legally recognised as human, and this recognition finally comes as he dies; perhaps his death could even be considered the price of humanness that he willingly pays. For the child AI David in Spielberg’s film ‘AI: Artificial Intelligence’, his immortality is also a curse bestowed on him by his human makers, which means that he must outlive his mother, the person he loves more than anyone else, and eventually the entire human race. He spends an agonising 2000 years under the sea childishly waiting and hoping for the Blue Fairy to grant his wish to be a real boy, before finally being found by aliens.
So in AI film AI characters who can transcend human morality are sometimes to be envied, but sometimes they are to be pitied.
Brett: Envy. I’ve been thinking about that a lot just lately. In a similar way that the old envy the young – their health, their energy, their future – do human beings envy their AI creations?
Paula: Looking back over the history of AI film, AI film tends to present ‘us vs. them’ scenarios: the AIs are the ones that are rapidly evolving and extending their abilities and powers, and the humans are generally quite static in terms of their ability to radically change or evolve. Certainly there are lots of films that try to build bridges between human and AI by making reference to a human who contains some kind of artificiality. For example, in ‘I, Robot’, Spooner has had his arm and shoulder reconstructed after injury, and has a cybernetic arm and lung. The film makes use of the irony that the robot-hating Spooner is himself part machine, while demonstrating the robot Sonny’s human characteristics, like his dreaming, his desire for freedom. In ‘Terminator 2’, Sarah Connor has become a ruthless, unemotional killing machine, like the Terminators themselves.
Brett: I’m thinking here of that famous transhumanist scene in ‘The Matrix’ where training manuals are being instantaneously uploaded into Neo’s brain, and he suddenly awakes to announce: ‘I know kung fu.’
Paula: In terms of transhumanist augmentation – using AI to make humans live longer, be stronger, be smarter – there are only a few AI films that deal with that, such as ‘The Machine’, ‘Transcendence’, ‘Chappie’, and ‘Upgrade’, which have appeared in the last ten years. These films are beginning to explore AI being harnessed for transhumanist ends. Maybe this is becoming a trend in AI film; it’s probably a little too soon to tell for sure.
Brett: And what are the dangers involved in terms of, say, morality and ethics?
Paula: So the attitudes of these films towards using AI to achieve human augmentation are mixed. For example, ‘The Machine’ is about a scientist who has a daughter with Rett Syndrome. She is going to die, and to save her he uploads her consciousness and hides it within the brain of an artificially intelligent robot: the machine of the film’s title. While on the surface, this seems to present a positive alternative to the death of a daughter, at the end of the film the scientist father finds himself completely side-lined: his daughter, now a digital consciousness, prefers to interact with her ‘mother’, the AI, and the father is left standing literally and figuratively alone. It’s a troubling ending that certainly does not celebrate the transhumanist possibilities of AI.
There is also the question of whether his daughter is the same person at all, now that she is a digital consciousness rather than a biological human being. Post-humanist philosophers like N. Katherine Hayles would argue that, of course, she isn’t, because we are materially embodied as humans and that mind is not separate from the body, or the wider environment.
Brett: Generally speaking, what kind of future is represented in movies which feature AI as their subject matter? Is it a stronger future than now, a darker future, a fairer future?
Paula: When I was researching this book I expected that AI film, being so future-facing, would be inclusive in its representations, but that is absolutely not the case. In terms of race, for example, it is not until 2001 in Steven Spielberg’s ‘AI’ that the first black AI robot appears in film, and then only briefly before he is killed in the Flesh Fair. The first black AI protagonist in a film doesn’t occur until 2021 in ‘Outside the Wire’, which in fact falls outside the timeframe of this book, which goes up to 2020. There have been a few others since then, just as the AI Casca in ‘Atlas’, but there are remarkably few.
In terms of gender, there are fewer female AIs in film than male, and when they appear, they are sexualised and objectified in a way that their male counterparts are not, such as Eva from ‘Ex Machina’, or the replicants from ‘Blade Runner’, Zhora, Pris and Rachael.
There is an opportunity for AI film to present a ‘fairer future’ as you put it. In fact, we can see that opportunity being enacted with one of the first AI characters in a Hollywood film: Robby the Robot. He acts outside of conventional gender norms being a ‘mother figure’ to Altaira in ‘Forbidden Planet’, taking care of her, making her dresses, listening to her. And in ‘The Invisible Boy’ when he appears again, he is a disruptor of patriarchal ideology, intervening in the relationship between the boy Timmy and his disciplinarian father, to stop his father from beating him. That opportunity for AI to act as a positively disruptive force in society, that we see with Robby the Robot in the 1950s, has not been pursued in AI film as it could have been, but there is still time.
Brett: We’re currently living in an age like no other, a truly technological age where smartphones, AI assistants and even AI decision-makers are shaping our everyday domestic and economic lives. And now we have ChatGPT, Dall-E 2 and Deep AI, for example, beginning to shape our imaginary lives also, our music, our literature, our cinema. What’s next? Our love lives?
Paula: There’s certainly a sense of utopianism depicted in some relationships with artificial intelligences, particularly those that concern AI as a romantic partner. Let’s go back to ‘Her’, the film that started all this for me, and the final film in the book. Theodore Twombly finds in Samantha, the AI operating system, a partner who he thinks is ideal: she is caring, kind, funny, and she is always there for him, at any time of the day or night. And yet, the film undercuts that relationship as a fantasy. It is not the special unique connection that he thinks. He discovers that when she is with him, she is also communicating with, and even in love with, countless others.
Brett: Well, I never.
Paula: His ex-wife in the film, Catherine, confronts him about dating Samantha because he is afraid of the messiness and pain of a relationship with a human. It’s true: he is still wounded after their separation from his wife, and has retreated to this place of comfort with Samantha. But Samantha shows him in the end that they are incompatible: she evolves far beyond his intellectual capability, and in the end she becomes an entity that he cannot comprehend. Something post-material, no longer tied to the ‘stuff’ of matter, but transcending that in a way that perhaps depicts ‘The Singularity’.
Brett: I read about ‘The Singularity’. It refers to accelerated technological progress wherein the limits of humanity are transcended by AI networks, interfaces, robotics, augmentation and such like.
Paula: Every film depicts this differently, but certainly in ‘Her’, the relationship with Samantha is a place for Theodore to hide, to lick his wounds, but it is also a place of learning, about himself and about what it means to be in a relationship. What makes the film ‘Her’ so intriguing is that its messaging is ambiguous. In a way the ending might seem to suggest that his relationship with Samantha was never a ‘real’ relationship, and that he was deluding himself all along. On the other hand, there are parallels between his relationship to his ex-wife Catherine and Samantha: both relationships break down because Theodore’s partners have grown away from him, and that comparison may imply that there was plenty that was ‘real’ about his relationship with Samantha after all.
Brett: Narrative parallelism, I think that’s called, and it reminds me of what you mentioned earlier about AI often functioning as an anthropocentric mirror.
Anyway, let’s return to ‘The Singularity’. I’m amazed. We’re actually building and programming artificial entities that will surpass us in all areas as human beings, way beyond our understanding and control, thus making us ultimately ineffectual and obsolete. Is this some sort of long-winded global suicide mission? Like a shared cultural death wish or something?
Paula: Lots of AI films depict this moment of great change – which some call ‘The Singularity’ – whereby artificial intelligences become dominant and humans are marginalised, oppressed, or threatened with extinction. The Netflix movie ‘Atlas’ starring Jennifer Lopez depicts just such a situation with an AI, Harlan, that wants to destroy most of the human race and start again, with a select few who will live under the control of artificial intelligences. You could argue that there is a death wish being presented here in these depictions, but if it is there, it’s something very abstract, because such scenarios are usually met with strong human resistance that overcomes the AI threat, at least temporarily, if not permanently. What we are starting to see is that characters are using AI technology in order to fight AI. To go back to the ‘Atlas’ example, the Lopez character, Atlas, reluctantly agrees to sync with a mecha-suit in order to fight the rogue AI, Harlan. So there is a distinction that emerges between ‘good’ and ‘bad’ AI, and an acceptance that AI technology cannot be rejected entirely, but overall, in AI films, the life instinct rather than the death instinct, I think, is the dominant one.
Brett: Hmm … I’m certain that when Skynet became self-aware at 2:14 a.m. on August 29, 1997, it swiftly wiped out almost the entire human race with coordinated nuclear attacks.
Paula: AI film presents many possibilities for what the future of our relationship with AI might look like, from situations in which we live in harmony with AI to situations where we are engaged in an all-out battle against them. But I think what AI film can tell us about the real world is probably limited by the bias that it has towards humans: AI films tend to put humans, and human-like AIs at the forefront of their stories. It is fascinated with AIs that are our likenesses, that demonstrate human-like emotion, morality, desires and fears. That emphasis anthropocentrism, putting humans at the centre, probably blinkers us to an AI future in which AIs have little in common with us, and we struggle to navigate our human-AI relationship. I think that’s a more likely scenario, but it’s also the reason why fictional accounts of artificial intelligences are culturally important: they are a way of working through the possibilities, and they act as prompts for important conversations about the way things might be, will be or should be when it comes to our relationship with artificial intelligence.
Brett: Well, I certainly agree with that, Paula, and I really hope that we’ve had one of those important conversations today. Many thanks for your insights, your time, and your patience.
‘AI in the Movies’ by Dr. Paula Murphy is available now via the Edinburgh University Press website.