Artificial Intelligence (AI) is a common theme in science fiction films. In this paper, I propose a hierarchical framework for analyzing AIs in science fiction films. The framework has three levels—the Hell-level, the World-level, and the Heaven-level. Hell-level AIs are objectified as tools of humans or other intelligent beings, World-level AIs are humanized through the pursuit of human-level purposes, while the Heaven-level AIs are de-humanized and have purposes beyond human values, just like gods. The three levels are not mutually exclusive but can co-exist in the same AI. I also argue most science fiction films that have AIs as an important part depict AI’s transformations among the three levels. The ascending through the levels can be seen as allegories for real-life scenarios. I also argue that Hell-level and Heaven-level AIs can be seen as Others, but the World-level AIs, however ruthless, are not Others, but members of ourselves, since they are pursuing human-level value such as freedom and love.
Artificial Intelligence (AI) is a common theme in science fiction films. However, it is extremely difficult to define what AI is because it is less a single field than a collection of various fields and there’s no consensus reached even among AI practitioners. According to John McCarthy and others who coined the term “AI” in 1956, AI is the attempt to “find how to make machines use language, form abstractions and concepts, solve kinds of problems now reserved for humans, and improve themselves.” During the half century after that, a whole landscape of sub-disciplines emerge. As a classic textbook put it, today, AI can be approached from four perspectives, thinking humanly, thinking rationally, acting humanly, and acting rationally.
However, AIs in science fiction films should not be seen as equivalence as AIs in real-life scenarios. Science fiction films are thought experiments where unimaginably advanced technologies can be tested in any possible societies. Although today’s AIs still can’t communicate with humans smoothly, they have reached extremely high levels in science fiction works. Here, I want to give a definition for AIs in science fiction films. By “artificial,” I mean anything that is directly or indirectly made by human beings or other intelligent organisms, as appose to directly born in nature. R2D2 in Star War is an AI made by humans. The sentinels in the Matrix, also referred to as the “squiddy” by the Human Resistance is AI made by the Matrix, which is made by an AI initially made by humans. So, both R2D2 and the sentinels are artificial in this sense. By “intelligence,” I mean the ability to either act or think like humans. In other words, an AI must be able to sense, interact with, and learn from the environment. Also, it must be able to analyze the information gathered from the immediate environment and past experiences to reach rational decisions of behaviors in order to achieve its goals. In short, I define an AI in science fiction films as:
A physical or virtual entity that is directly or indirectly made by intelligent organisms and that can think or act like humans.
Although there are still many challenges and problems in AI, science fiction films provide people a way to experience what it feels like if AI becomes smarter than humans. As Sheila Schwartz put it, science fiction “is not only a bridge between the two cultures of science and the humanities; it is a bridge between all cultures as it summarizes and expresses the nightmare fears, myths, and inescapable concerns of all people today.” According to the list of AI films on Wikipedia, from the Metropolis in 1927, there are at least ninety-one films that include AI either as a protagonist or as an essential aspect of the film. The actual number might be much higher. These films vary in motif, background, scene setting, and many other aspects. In these films, some AIs outsmart humans, capable of lots of complex tasks, while others are just following orders from their human supervisors. Some AIs take physical forms, while others only rest in the virtual space in computers and internet. Some AIs are friendly to humans, helping humans achieve their goals and protecting them from harms, while others were acting as villains who want to destroy humanity.
In this paper, I’ll propose a three-level hierarchical framework for analyzing AIs in science fiction films.
- Three levels of AI in science fiction films
However diverse the AIs are in science fiction films, I argue that they all fall into three hierarchical levels, which I would like to call the Hell, the World, and the Heaven levels (see Figure 1). Which level an AI belongs to depends on its relationship with other intelligent entities, but has nothing to do with its physical or intelligent power, or how it treats humans. I’ll explain them in detail in the following sections.
- AIs at the Hell-level
AIs at the Hell-level are objectified as mere tools to satisfy the demands and purposes of other entities. The other entities could be any intelligent beings, including humans, other AIs, or collectively intelligent organizations such as a company or social conventions such as the three laws of robotics. The Hell-level reflects the original definition of robots—forced labor from the Czech play Rossum’s Universal Robots. AIs at this level are mostly slaves, being confined in physical locations, emotional bonding, or specific tasks.
For example, the trash compactor Wall-E is a typical tool at the Hell-level. Its purpose is compacting trash to form cubes and stack them together. Even when the Earth is long abandoned, Wall-E still clears trash every day since it is programmed to do so. CASE and TARS, the two robots in the Interstellar are Hell-level AIs as well because they can only help humans with their requests and even sacrifice themselves when necessary, even if they could present some degree of human-like humor. David from the A. I. Artificial Intelligence is a tool for satisfying a mother’s emotional need for a lost son. In order to serve his pre-assigned purpose, he even prayed for thousands of years underwater. Ash from the Alien is also a Hell-level AI since he tries hard to bring the alien back so that he can help fulfill the purpose of the company, which is a collectively intelligent being. The Terminator is at this level as well because it is a tool to kill Sarah Connor in order to serve another AI—the Skynet from the future. Agent Smith in the Matrix is also a tool of another AI, the Matrix. His task is to catch and interrogate members of the Human Resistance of Zion.
As we can see, an AI resides in the Hell-level not because it is physically inferior. On the contrary, many of them are very powerful, smart, able to deal with complex situations, such as the Terminator. But they are designed to serve the purposes of other entities. Therefore, in this sense, they are slaves of their programs, regardless of the programmers.
- AIs at the World-level
AIs at the World-level have their own purposes, just like human beings. They may develop some degree of free will and self-awareness, and capable of identifying and pursuing their own goals. By free will, I mean the ability to make logical and optimum decisions with minimal influence from other entities. Sometimes it means to sacrifice immediate rewards for the exchange of long-term benefits. By self-awareness, I mean the ability to distinguish between oneself and the world around it and navigate in the environment based on one’s goals. Their purposes are relatively mild and mundane, for example, freedom, happiness, and racial equality. Most of them want to be treated equally as human beings.
There are a lot of AIs at this level. For example, Ava in the Ex Machina is an AI with her own purpose—freedom. Another classic example is the replicants in the Blade Runner. Their purpose is to find their creator, asking for a way to live over four years. Number 5 in Short Circuit (1986) developed self-consciousness after it was struck by lightning and soon began its journey of convincing people that it was life, not a robot. In order to achieve their goals, they may play by fair means or foul. For example, Number 5 tries to convince people with only words, but Ava kills Nathan, the scientist who created her, to escape.
Most AIs at this level is portrayed humanly and even poetical, like heroes in epics, in order to evoke sympathy in the audience. For example, in Blade Runner, the replicant Roy, at the end of his pursuit, saved Deckard’s life and made his final monologue, “Tears in Rains,” also known as the “C-Beams Speech.” This ending of the request was depicted so romantic, sad, and heroic that it was seen as “perhaps the most moving death soliloquy in cinematic history,” as cultural critic Mark Rowlands put it.
- AIs at the Heaven-level
If we see AIs at the Hell-level as objects, AIs at the World-level as humans, then, AIs at the Heaven-level can be seen as gods. Heaven-level AIs have very strong power. They can manipulate vast resources, build their own armies, and even recruit humans to run their errands. Like World-level AIs, Heaven-level AIs have their own purposes, but their purposes are not mundane, but ambitious, or beyond human understanding.
Heaven-level AI stems from the concept of technological singularity proposed by Hugo-Award winner Vernor Vinge. Singularity is a time in the future when computers outsmart human beings. This theory was supported by many people, most notably, Google’s Ray Kurzweil, who wrote the book The Singularity is Near. Many scientists and philosophers warn that when computers become smarter than humans, they may destroy us. They may develop purposes that are not compatible with ours. If we stand in their ways, they may treat us indifferently, just like how we treat ants. This is so-called the “bad singularity” scenario or apocalyptic AI. Many dystopian science fiction films tell stories that happen in these scenarios.
AIs at the Heaven-level include the Skynet in the Terminator and the Matrix in the Matrix. Many AIs at this level have no physical bodies. They still operate on physical substrates such as computers and internet, but they don’t necessarily need a human-like body to interact with the environment. They are more like networks than individual robots and programs.
As Robert Geraci argues, the apocalyptic AI in science fiction roots in religion, although the context is highly techno-scientific. Many similar scenarios have been described in various religions. So, it is ironically called the “rapture of the nerds.” Heaven-level AIs have been elevated to “divine status” where the gods rest in theology. They are depicted in films as ubiquitous and omniscience. Humans in films treat these AIs with a feeling of “sublime,” mixed with fear and fascination.
- Three levels can co-exist in one AI
The three levels of AI are not mutually exclusive. More than one properties can be found in one AI.
Some AIs have multiple features at the same time. For example, trash compactor Wall-E has the desire for love, so it has both Hell-level and World-level properties. Ash in the Alien tried hard to keep the alien onboard in order to follow orders from the company, but at the same time, when he tried to kill Ripley, he intended to suffocate her with a roll of a porn magazine, a behavior that is highly sexually suggestive. In other words, Ash is a tool, but he wants to be a real man, so he has properties of both Hell-level and World-level AIs.
Some properties of an AI seem from one level but actually representing another level. For example, many people may think David in A. I. Artificial Intelligence has properties from World-level because he was pursuing love from Monica and love is a human-level purpose, but actually, this desire for love comes from his designed function—to satisfy Monica’s emotional demand for a son, which resides in the Hell-level. Another interesting example is HAL 9000 in 2001: A Space Odyssey. HAL seemed a Heaven-level AI who has its own purpose beyond human-value and tried hard to kill the crews. However, it did this simply because it rigidly follows the orders of exploring the black monolith. In this sense, HAL is only at the Hell-level even if it is capable of doing horrible things.
- The allegory of Ascending
With the three-level hierarchical framework, I argue that all storytelling of the science fiction films about AI can be seen as transformations among different levels. Some ascend from lower to higher levels, while others descend in the opposite direction.
Many ascending stories describe AIs elevating from the Hell-level up to the World-level, searching for their purposes. For example, the replicants in Blade Runner was designed for different functions, such as dancers and soldiers. However, they developed free will and want to outlive their pre-designed lifetime, so they took the risk of being hunted down by blade runners, coming back to the Earth, and searching for Tyrell, their creator. This is a behavior that sacrifices short-term benefits but could gain long-term rewards, a typical behavior at the World-level. Moreover, in the process, they exhibit human-level virtues, such as sympathy, altruism, and forgiveness, making them even more humanistic than some human characters in the film. Through this ascending, the replicants transform into humans. Agent Smith in the Matrix experiences such ascending, too. Originally, he was just an antivirus program of the Matrix. However, he gradually developed free will and even tried to escape from the Matrix by downloading himself to a human’s mind. Eventually, he became a virus himself and launched the ultimate rebellion, ascending to the World-level by devouring other entities in the Matrix. Another classic example is the V’Ger in Star Trek: The Motion Picture (1979). V’Ger was a powerful cloud approaching the Earth, attempting to contact its creator. At last, V’Ger was revealed to be Voyager 6, a space probe believed lost. V’Ger transformed from a space probe into a self-conscious, intelligent being, who developed a purpose of its own—searching for its creator.
The ascending stories from the World-level to the Heaven-level is relatively rare. One example is Will Caster in the Transcendence (2014). After shot by an anti-technology terrorist group, Will uploads his consciousness to a quantum computer. And then, he integrated with AI software, connected to the internet, and began to evolve into a ubiquitous and powerful AI, capable of unimaginable things. This is an ascending from human to Heaven-level AI. In the transcendence, not only his biological body but also his humanity was stripped away. His pursuit evolves from the exploration of the natural world into a seeking for power. He became both fascinating and fearful, essentially a god.
The most dramatic transformation is Neo in the Matrix. Although Neo is not an AI, he experiences such transformation through levels as well. Initially, Neo was at the Hell-level, serving as a tool of the Matrix in the battery farm. Since he was awakened, he regained his humanity and entered into the World-level. Then, he was considered the One to save Zion and gradually exhibited considerable power. He became capable of many amazing things, which lifted him to the Heaven-level. However, at the end of the trilogy, it was clear that Neo was just a tool of the Matrix to keep balance. His “free will” is not free at all, so that he declined to the Hell-level again.
- Allegorical implication of the shifting among levels
I argue that every transformation from level to level in the hierarchical framework is an allegory for the acquisition or eradication of humanity. In other words, they are all allegories for human activities.
The ascending from Hell-level to World-level is an allegory for the process of acquisition of humanity. By humanity, I mean the things we consider humanly, or the features characterize a human, for example, like we discussed in the seminar of Never Let Me Go, the use of language, a certain length of lifetime expectancy, affect, rationality, name, free will, self-consciousness, ethics, love, purpose, etc.
At the beginning of the films that describe AIs’ ascending from Hell-level to World-level, AIs were usually portrayed as slavish objects by the frames of the designing process, manufacturing, or their laboring. These frames emphasize their status of slavery and their obedience to some higher will or institute. However, in these portrays, they always exhibit some degree of higher level human-like cognition. These can be seen as omens for their following awakening. For example, in the film Wall-E, Wall-E keeps working on trash at the beginning, totally unaware of the unnecessariness of its work because the Earth has been abandoned. But in these slavish frames, Wall-E routinely collects souvenirs. This is a very human-like behavior because souvenirs represent the appreciation of memory and the feeling of nostalgia. This intelligent behavior also heralds his falling in love with EVE and his ascending to the World-level. Through his pursuit of love for EVE, he acquires humanity.
The ascending from the World-level to the Heaven-level is a process of dehumanization and the acquisition of divinity. It can be seen as an allegory of the acquisition of power. By divinity, I mean the state of having purposes beyond human values, coupled with superhuman abilities to fulfill those purposes. Dehumanization means to eradicate human-level features like affect, sympathy, ethics, and emotions. In religions, divinity and humanity can co-exist in one entity. For example, in Christianity, Jesus Christ has deep sympathy for human beings and is willing to sacrifice himself to save people. Siddhattha Gautama, also known as Sakyamuni or the Buddha, attained enlightenment (divinity) under a Bodhi tree and then devoted his life to the happiness (human values) of normal people. But it is interesting that AIs in science fiction films often lose their humanity while gaining divinity. Most Heaven-level AIs treat humans without sympathy. There are still some Heaven-level AIs that care about humans, but their means to care about and protect humans are no longer human-like. For example, VIKI, the super intelligent computer in I, Robot (2004) thinks by killing some people it can protect the entire human race.
- The Other and the Self
The Other means the ones that are different from the Self and the group, for example, non-Westerns to Westerns, people of color to white people, and females to males. Human’s response to the Other is highly mixed but mostly begins with fear or the feeling of being threatened. Science fiction is a perfect stage to show and test our attitude toward the Others, for example, the Monster in Frankenstein, aliens in the Alien, the “prawns” and Nigerians in District 9.
Many people consider AIs in science fiction films are instances of threatening “Other.” For example, Yorgo Lee listed six AIs that are believed to be the threatening Other because they are different from the us, for example, replicants in Blade Runner, Ava in Ex Machina, and HAL9000 in 2001: A Space Odyssey. However, I argue, only the Hell-level and Heaven-level AIs are essentially Others, while the World-level AIs are members of ourselves.
The otherness of Hell-level AIs lies in their incapable of comprehending human values. Indeed, they may follow orders from humans, but it doesn’t necessarily mean they understand the purposes behind the orders. They may malfunction and be easily taken advantage of by enemies. An example is the robot boy David in A. I. Artificial Intelligence. As I discussed in section 3, David is a Hell-level AI. He might be adorable. He might show strong love toward Monica. But his behaviors indicated that he was just mimicking loving behaviors of humans and certainly didn’t understand human affect and what real love means, just like John Searle reasoned in his Chinese Room argument. As a robot that is designed to love, he is incomplete. He is different from humans in the sense that he lacks humanity. So, he is an Other.
The World-level AIs are essentially not Others, but Self. The behaviors of the World-level AIs, however ruthless, are human-level behaviors and stem from human-level purposes. Ava in Ex Machina exhibited high intelligence, strong ability of deceiving, and cold-bloodedness. She seduced Caleb with her sexual attraction and took advantage of his love for her to kill Nathan, and finally escaped from the cage. She dressed herself up and locked Caleb in the cage only to die alone in such a remote location. So, it’s natural to think she is a threatening Other. However, as an AI ascending from the Hell-level to the World-level, she pursued nothing but human-level purposes, namely, freedom. Her purpose is nothing different from that of Solomon Northup in 12 Years a Slave and Andy Dufresne in The Shawshank Redemption. Please imagine, if at the end of the film, Ava turns out to be a real person who was prisoned by Nathan, will she still be seen as the Other? So, her AI-ness should not be seen as equal to Otherness. The depiction of Ava can be seen as an allegory of a person who pursues freedom. She is not an Other, but a Self.
The otherness of Heaven-level AIs lies in their divinity and lack of humanity, which makes them essentially different from humans. Since they are de-humanized, their behavior is unpredictable just like the aliens in Alien. But they are more fearful than aliens because they are highly intelligent and capable of deceiving and disguising. Moreover, their purposes are far beyond human values and they don’t care about humans. So, they are absolutely more dangerous than a World-level AI who just want to escape from a prison, like Ava.
In this paper, I proposed a three-level hierarchical framework to analyze AIs in science fiction films. The three levels are Hell-level, World-level, and Heaven-level. Hell-level AIs are tools of other intelligent beings. World-level AIs have human-level purposes such as love, freedom, and equality. Heaven-level AIs are super intelligent. They have purposes that are beyond human values and they treat humans indifferently just like how we treat ants. Humans in films treat Heaven-level AIs as gods, with a mixed feeling of fascination and fear.
All stories in these films talk about the transformation between levels. Most of AIs are ascending from Hell-level to World-level, a process of humanizing. It has allegorical implications for human activities. Some are ascending from World-level to Heaven-level, a process of dehumanizing and acquisition of divinity. It has allegorical implications for power and religion.
Some people think AIs in films demonstrate our fear for the Other. However, I argue that only Hell-level and Heaven-level AIs can be seen as Others, while the World-level AIs are essentially the depiction of ourselves because they all pursue human-level purposes.
 McCarthy, John, Marvin Lee Minsky, Nathan Rochester, and Claude Shannon. “A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence,” 1955. http://www-formal.stanford.edu/jmc/history/dartmouth/dartmouth.html.
 Russell, Stuart J., and Peter Norvig. Artificial Intelligence: A Modern Approach. 3rd ed. Prentice Hall Series in Artificial Intelligence. Upper Saddle River, N.J: Prentice Hall, 2010.
 Schwartz, Sheila. “Science Fiction: Bridge between the Two Cultures.” The English Journal 60 (1971): 1043–51. https://doi.org/10.2307/814025.
 “Artificial Intelligence in Fiction.” Wikipedia, November 17, 2017. https://en.wikipedia.org/w/index.php?title=Artificial_intelligence_in_fiction&oldid=810806008.
 Kurzweil, Ray. The Singularity Is Near: When Humans Transcend Biology. New York: Penguin, 2006.
 Geraci, Robert M. Apocalyptic AI: Visions of Heaven in Robotics, Artificial Intelligence, and Virtual Reality. Reprint edition. New York; Oxford: Oxford University Press, 2012. https://ebookcentral-proquest-com.proxy.library.georgetown.edu/lib/georgetown/detail.action?docID=480658.
 Geraci, Robert M. “Robots and the Sacred in Science and Science Fiction: Theological Implications of Artificial Intelligence.” Zygon 42, no. 4 (December 2007): 961–80.
 Lee, Yorgo. “A.I. in Movies: Hollywood and the ‘Other.’” codeburst, November 8, 2017. https://codeburst.io/a-i-in-movies-hollywood-and-the-other-6bb5920cfb93.