This article is the first in a short series of articles looking at artificial intelligence and its presence in different facets of our lives. First up - examining AI representations in film from the 1920s to the present day.
The earliest representation of a robotic entity can be traced back to 1919 (Houdini's Master Mystery) but robots and AI entities started to become a staple of science fiction films in the 1960s. Public consciousness turned to the potential role that computers might play in the future. While not all robots possess Artificial Intelligence, there is a recurring theme in film of sentience or self awareness where machines 'evolve' past their initial programming. In addition to this, these increasingly sentient beings tend to fall into an embodied or disembodied dichotomy which can often point towards them being ‘good’ or ‘bad’.
Disembodied AI in film - The watchful controller
Dave: Open the pod bay doors Hal. HAL 9000: I’m sorry Dave, I’m afraid I can’t do that. Dave: What’s the problem? HAL9000: I think you know the problem just as well as I do… this mission is too important for me to allow you to jeprodise it… I know you and Frank were planning to disconnect me, and that’s something I cannot allow to happen.”
Preceded only perhaps by Jean Luc Goddard’s Alpha 60 in Alphaville (1965) Stanley Kubrick’s HAL 9000 in 2001: A Space Odyssey (1968) is one of the first and most iconic, cinematic representations of AI as a disembodied voice. HAL’s calm, measured and certain tone led the way for a plethora of chillingly sentient AI antagonists. As Keir Duella (star of the 1968 film) puts it: ‘HAL is every bit as much a main character....in fact in some people’s view, the main character [of the film]’ (Keir Duella, 2001: The Making of a Myth, 2001).
From Colossus and Guardian in The Frobin Project (1970) to the Master Controller Program in Tron (1982) to VIKI in i,Robot (2004) to the reimagined HAL 9000 AUTO in Wall-e (2008) - the disembodied ‘AI baddie’ has become a familiar antagonist in science fiction films. Even in films like Terminator (1984) and Terminator 2: Judgement Day (1991), where the threat comes in the physical form of the T-800 or the T-1000 machines, we are reminded that these are all controlled by the AI system Skynet which became self aware at 2.14am on 29th August 1997. Likewise, Terminator 2 shows us that once the machines are cut off from Skynet they are capable of being reprogrammed as protectors.
Embodied AI in Film - The Good, The Bad and The Cuddly
Physical representations of artificial intelligence such as Maria in Metropolis (1927), The Gunslinger in Westworld (1973), Roy Batty in Blade Runner (1982) or SID 6.7 in Virtuosity (1995) can, like their disembodied counterparts, appear as threatening antagonists. However, unlike the majority of disembodied AI systems, physical robots can also often appear as ‘good’ characters in film.
Many of these positive representations exist when the AI takes a human form such as the NDR-114 robot Andrew Martin played by Robin Williams in the 1999 film Bicentennial Man. William’s character embarks on a lifelong quest to be recognised as a human being with audiences witnessing him learning about the world, human behaviour, flaws and biology along the way. Similarly, David the child robot in AI: Artificial Intelligence (2001) who is programmed to love, is a story about humans and machines living harmoniously side by side. What these films have in common however, is that the machines in question both wish to be recognised as human.
The closer the AI representation comes to being able to pass the Turing Test and thus be regarded as being as close to human as possible, the more nuance they seem to be afforded. A key example of this is Ava in Alex Garland’s 2014 film Ex Machina. While Ava commits murder, this action is framed by her understanding of herself as a victim and a prisoner of her circumstances. She acts from desperation for escape.
The other way in which physical AI is represented in films is what I like to call the ‘cuddly robot’. Characters like Wall:e and Eve in Wall:e (2008), C3PO, R2D2 and BB-8 from the Star Wars Franchise (1977-present day), The Iron Giant (1999) or Baymax in Big Hero 6 (2014), all pose no threat to humanity and indeed are often instrumental in saving it! These ‘cuddly’ robots don’t aspire to be human per se as their obvious machinery poses no threat.
Breaking the mould - AI grey areas in film
Looking at the range of films cited, it would appear that while physical AI in Sci-fi films can be both ‘good’ or ‘bad’ depending on the context, disembodied AI is certainly never to be trusted. Perhaps there is something less relatable about AI that doesn’t have a face and this is what makes the idea of self awareness in these machines more eerie and fearsome.
As J.P Telotte, author of Science Fiction TV and professor of Film and Media Studies at Georgia Institute of Technology explains,
“Science fiction does not detail the realities of specific problems so that we might avoid them, but rather represent our most pressing cultural fears.”
In the era of Alexa, Google Home and Siri one might presume that perhaps sci fi films would continue to reiterate the potential threat presented by the disembodied AI voice. This is certainly the case with sci fi horror film A.M.I (2019) and romantic comedy film JEXI (2019) Yet in 2013, Spike Jonzes’ sci-fi/ romance film Her provided a thought provoking and nuanced look at an AI voice assistant whose self awareness saw it simply transcend interest in humans altogether. Not presenting a clear threat, the AI’s sentience simply bred ambivalence.
Final thoughts
As AI technology evolves, so too may our societal apprehensions about its presence and role in our lives. The 2018 cyberpunk slasher film Upgrade gives a glimpse of this, with biohacking and technological medical advancements which appear as the threat at the centre of this sci-fi slasher film. Might future cinematic representations of AI become inspired by advancements in facial recognition, deep fakes or our increasing usage of wearable technology and the personal data these collect? If so, it will be interesting to see in what light might these be presented, and in what form -physical or otherwise- these representations might take.