It can sometimes happen that --- as a historical era becomes sufficiently aware of the scientific theory of its times --- it responds by internalising the broader points of debate. In the process, the popular imagination of the time might well seize upon the immediately attractive or disturbing aspects of scientific discourse without necessarily concerning itself with the finer points of discussion. This selective extrapolation is a dominant feature, for example, of late nineteenth-century populist speculation. It lights upon the hectic features of the work of Alfred Lyell and Charles Darwin --- its rejection of catastrophism, its embrace of evolution --- and applies them to pressing concerns of the time, such as race, empire and class.  ‘Given the dinner-table wisdom that genteel Anglo-Saxons towered above their black butlers, evolution cast a shadow over ancestral purity. Lyell himself was irresistibly drawn to the theme: go back umpteen generations and would blacks and whites find a common ancestor? Itself the descendant of an ape?’ (Desmond and Moore, 190)(3) As the extract suggests, the mind of any historical period tends to skip many steps, and to leap over theoretical minutiae to fasten upon the broad features of debate. In this case, the Victorian imagination selects the distinguishing feature of the findings of Darwin and Lyell --- the competitive aspect of evolution --- and applies it, first as metaphor and then as fact not to the category of species but to that of race. As a result ‘Darwinism’ --- a useful if inaccurate allegory of the mind --- acquires a colloquial but deeply discriminatory racial dimension far away from the theoretical field that Darwin sought to establish. The late nineteenth century in England and the United States thus projects its intensely human hopes and fears onto the evolutionary debate. It uses Darwinism as a mode through which to understand the prevailing crisis of thought of the time. The moral allegory that develops out of such a situation frames its own core question. Does the survival of the fittest constitute the survival of the physically or morally robust? The question is nowhere near the research triggers that motivated Darwin, but re-emerges whenever a subsequent age recalls Darwinism.

In the essay that follows I argue that something of the same process overtakes the discussion around Artificial Intelligence or AI. The early twenty-first century picks up this legacy of the twentieth century and endows it with its highest aspiration and deepest terror.  The great hope of the period is to seek human perfectibility that always seems to be just beyond its reach. The other face of this hope is the great fear that humanity may prove to be redundant. This is the debate that --- in an effort to understand and shape its destiny --- the millennium projects on to AI. Cinema serves as the vehicle of choice that mobilizes this debate in the popular imagination. To get the argument off the ground however, it is necessary to map the disciplinary space that AI occupies. The section that follows will attempt to do so.

1 Mapping the field
1.1 Definitions of AI

It is helpful to track the development of AI by thinking about the steadily inward-turning shift in the definitions that seek to map the field. The most famous of these early formulations is the query ‘Can machines [computers] think?’ that history attributes to Alan Turing, one of the founders of the discipline. (11) (Persson 88) This apparently banal question actually pushes its hearer to conceptualize AI as a distinctly inferior version of human intelligence, with specific reference to the faculties of cognition and problem-solving.  Another early formulation is John McCarthy’s description of the task of ‘making a machine behave in ways that would be described as intelligent if a human were so behaving.’ (Kavanagh 13) (7)Note the way in which McCarthy, a founder of the discipline, kennels AI as a sufficiently good watchdog that might almost be human. Indeed, one of the problematic legacies of this period is the condescending evocation of the instrumentality of AI. Formulations of the period see AI as an agency that its human designers can use for their purpose and shut down at will. The scientific community sees AI at this point as an adjunct of human purpose, fit to ‘deal with tasks that would otherwise require human intelligence and intervention to be performed successfully.’ (Kavanagh 13)

Two generations later, however, there is a decisive shift in the conceptualization of AI. As its cognitive reach and scope broaden, AI seems to position ‘a new ecosystem to the world.’ (Kasapoglu and Kurdemir 1) Where AI once served humanity by enhancing its decision-making abilities it now begins to mime the functions of human intelligence.  The question that emerges concerns the nature and extent of the role to be played by humanity when ‘AI algorithms and autonomous systems [can take decisions] without a human in the loop.’ (Kasapoglu and Kurdemir 5) It is clear that the phase that studied the instrumentality of AI --- as a research assistant to human intelligence --- is a thing of the past. The present age deals with the way in which AI may now reveal human intelligence to be redundant, a thing of no value. As humanity struggles to remain relevant in the face of this emerging challenge from AI, it articulates twin fears. The first is the primeval fear that AI will simply replace human skills, and ultimately render the race and its achievements obsolete. The second is the more sophisticated fear that the decision-making faculty that AI demonstrates --- that derives from its ability to strategize data-banks through algorithms of management --- may render democracy vulnerable to ‘networked feudalism’ (14)(Unver 127) or dictatorial governance manipulated by those who control systems that AI sustains. When articulated in this way by theorists, AI seems to be sufficiently incomprehensible to the great majority of people as to carry no material threat. When the popular imagination restates the case in cinematic terms however, the situation plays out very differently, as the next section will try to show.

1.2 Cinema and AI
AI begins at a disadvantage in the domain of cinematic application rather as it does in the domain of theoretical discourse. It seems to open up a new range of tools to produce computer-generated imagery that dazzles to disappoint. The initial use of AI to generate special effects in SF films again draws attention to its instrumentality, its ability to operate as a support to the major cinematic argument rather than generate debate on its own terms. Indeed, one might argue that a dependence on such special effects weakens the premises of cinema itself, leading to accusations of ‘impoverishment of narrative and depth.’ (Abbott 89) This dependence on computer-generated imagery that characterizes certain kinds of science fiction in cinema again adds to the sense that AI is somehow a sensational and cheapening force in film. It seems to sensationalize content and audience-response with minimum effort and maximum effect. Later efforts to retro-fit the cinematic deployment of AI with new-found philosophy do not always work. Even if one argues that the imagery so generated allows viewers to imagine worlds of the future so that ‘the medium [is] the message,’ (Abbott 89) it seems an incomplete conceptualization of the possibilities of what AI and cinema can do together. 

A more recent effort to think through this field suggests that AI plays a critical role in actualizing the way in which the human and the non-human come together to produce a new kind of knowledge. ‘The stacks of animation in AI constellates another point in animation’s history of the ways in which humans and nonhumans are suffused or intertwined in its constitution, even when seemingly separate from a distance...’ (Johnston 162) (6) To consider the role of AI in knowledge-production in such a way may be more elevating than to speculate on its ability to generate fields of images. At the same time, this seems to damn it with faint praise. There has to be a more robust way to study the way in which AI and cinema relate to each other, if only because the reverse side of that hope, namely fear, is so strong. Indeed, it might be said to be at the heart of the current strike in Hollywood that engulfs the entire film fraternity. Articulating the threat that AI poses to cinema, Duncan Crabree-Ireland --- the negotiator for the chief trade guild SAG AFTRA --- explains the matter in this way. ‘The [management] says that our background performers should be able to be scanned [using AI] , get one day’s pay and their companies should be able to own that scan, their image, their likeness, and should be able to use it for the rest of eternity on any project they want, with no consent and no compensation. So if you think that’s a ground-breaking proposal, I suggest you think again.’ (12)(Pulver and Shoard, https://www.theguardian.com/culture/2023/jul/14/the-hollywood-actors-strike-everything-you-need-to-know We cannot wish away the elemental fear that people in the film industry experience when they think that --- owing to AI --- their human presence and identity might be at the mercy of the industry that can replicate these features throughout time and space. The grimly parodic echoes of the Creation narrative from Genesis --- ‘image,’ ‘likeness,’ ‘eternity,’ --- are deliberate and paralysing. AI seems to stand forth as a satanic creator. It appears to generate a race that feeds off human identity, and then proceeds to render it of no value.  With this contemporary protest, the wheel comes full circle. AI began its troubled journey as a seemingly inferior supplement to human ability. It now stands forth as a resilient and adaptable alternative that may well dispense with its former creator.

However melodramatic such a statement might seem, we have to accept that this formulation is at the centre of an unprecedented controversy at the heart of the cinematic world. Hollywood speaks its anguish at the way in which AI can make the human world of little consequence. Indeed, it practically casts AI as Cain, in its ability to erase human life and identity. In consequence, AI seems bound to wander homeless through a world that is largely of its creation. In the face of this concern, on what terms can one recover the value that inheres in the relationship between AI and cinema?

2 AI and the cinematic multiverse
2.1 A complicated common ground
An emerging popular space in cinema --- the multiverse --- seems hospitable to a particular set of applications of AI. I should hasten to add however, that while this kind of cinema relies heavily upon AI and augmented reality to construct its conflicting topoi, people its worlds and carry forward its message, this is not the kind of relationship that I wish to signal. These features are very much the external markers of the instrumentality of AI. I move on instead to the philosophical dimensions of AI that the multiverse seems to effortlessly internalize and require. These constitute the broad imperatives behind AI. As such, they include its determination to reach out to perfectibility for the world of the future, its ability to endlessly alter and re-imagine characterisation, and reconfigure both space and time-management. However, before developing this line of argument that relates AI to the cinematic multiverse, it is necessary to identify the characteristics of the multiverse, and to speculate on the nature of the reality that it offers.

The most obvious visual and metaphysical quality of the multiverse is the way in which it organizes community-based storyworlds that both collaborate and compete with each other. These often take their rise from the comic-book fiction but --- supported by evolving franchises and studio networks --- rapidly outgrow their beginnings. The apparent diversity of life-forms, themes and landscapes that mark these storyworlds is perhaps the first feature that the viewer notices. As the following analysis suggests, the components of a multiverse, even in comic-book form are disturbingly difficult to follow. ‘In principle these storyworlds can be viewed as counterfactuals; changing particular elements in the characters’ situations, they relate to one another as “what if” versions. But because a baseline reality is often difficult to discern within this constellation of worlds, the multiverse poses considerable processing challenges. (Kukkonen 55) (10)  While Kukkonen goes on to suggest that the use of costume and the judicious positioning of characters who serve as guides or surrogates for the reader are critical to the comic-book multiverse, the cinematic multiverse goes further. This however, belongs to a later section of the argument. The point at issue here is that the broad purpose of all these storyworlds is to offer possibilities of perfectibility --- of form, action and impulse --- that sometimes come together and at other times compete with each other to offer the viewer a set of alternatives. Each world is self-contained, and yet each relates to the other in terms of common characters, plot lines and imagery. The multiverse reaches out to perfection, offers algorithms of choice, and speculates on the possibility of intervening in a historical time-line and altering it for the better. These are very much the metaphysical imperatives that drive AI in its conceptualization and that is why it is possible to look at the cinematic multiverse in part as an allegorical representation of the speculative impulses that shape AI in its philosophical dimension. Among the different options on offer, the Marvel multiverse --- as it translates from page to screen --- is the most flexible alternative to study. The first reason is the nature of the creative autonomy it offers the viewer. It repositions the viewer as ‘an active participant in the generation of the fictional heterocosm.’ (14)(Silvio 41) The second is that --- undergirding its seeming visual variety --- there is a strong thematic and organizational unity that holds the Marvel multiverse together, in a seamless process of ‘continuous relativization and revision, a process which is both collateral and interminable.’ (Silvio 43) The section that follows will debate the extent to which this inherited wisdom concerning the world of Marvel comic book fiction holds good when translated into cinematic terms. The specific points under consideration will be the ways in which they internalise and deploy the foundational principles of AI: its thematic reach for perfectibility, its visual reach for choice and its ability to intervene in history and alter its course.

2.2 The Marvel Multiverse
The first point that emerges in the context of the Marvel multiverse is that the assemblages or storyworlds that cinema constructs take their rise from a precise historical context: the impact of 9/11 on the United States so as to not merely humble the nation but also to lay bare its vulnerabilities to the world. As one of the raft of Marvel superheroes explains, ‘The idea was to bring together a group of remarkable people, see if they could become something more... to fight the battles we never could.’ (Pardy 103) Just as AI begins on a compensatory note --- of machines stepping up to accomplish human directions after the Second World War --- superheroism on screen takes a fresh turn after 9/ 11 by seeking to push forward when human heroism begins to pull back. The two most recent films to both contextualise and re-examine the imperatives of superheroism within the Marvel multiverse that this paper will examine are Ant Man and the Wasp: Quantumania (released February 2023) and Guardians of the Galaxy: Vol 3 (released April 2023). Both market themselves as the concluding films in their sequences. While this is not as yet verifiable, it is fair to say that such a sense of an ending allows both films to look back on their prequels and also on the lines of argument that conclude within the films. This in turn opens up a certain space for speculation and account-reckoning. It is also reasonable to assume that they will indicate lines of inconclusiveness as well, and think about themes and ideas that perhaps resist finite closure. These possibilities may resonate in the larger world of critical thought that encloses cinema as well, but to take these ideas forward it is necessary to engage directly with both films.

2.3 Ant Man and the Wasp: Quantumania (13)
The film begins on a note of disrupted expectations as regards superheroism and scale. The opening sequence in the Quantum Realm that serves as a prelude shows a surprising moment of mistaken compassion. Janet Van Dyne helps Kang to escape. The viewer does not as yet know the consequences of this moment but a series of episodes conveys the idea that superheroism needs to include a measure of humane thought and action. In a succeeding sequence --- this time on earth --- the viewer sees Scott Lang reading from his bestselling memoir Look out for the Little Guy that constitutes another nudge in the direction of a re-imagined heroism. Lang’s daughter Cassie organizes a protest to rehabilitate homeless people and Hank Pym (Lang’s associate) repeatedly reminds viewers to focus on the ‘little guy.’ On the face of it these frequent allusions to the idea that micro-change repays attention suggests a crucial redefinition of superheroism. It seems appropriate that Pym (the creator of the Ant Man suit) should be an entomologist and that Lang (who wears the suit) should repeatedly formulate the same thought. In reality though, this argument plays out in a remarkably sophisticated way. At a turning-point in the film, ant swarms come together to help this fractured family escape from the Quantum Realm. This is impossible unless the swarms cooperate with each other to form a pyramid that will help take the protagonists to safety. As the ant-swarms come together, it becomes clear that the domain of once despised computer generated (1) imagery --- thought to be the sole benefit of AI in cinema --- is now changing out of recognition. It is not just the visualisation of the thought that salvation is possible if cooperation across species can work. It is the actualization of this thought as well. (2) The concept of the biological swarm as an image of cooperation facilitates the development of the robotic swarm of networked operations that is at the heart of AI. ‘At the epicentre of swarm-robotics is self-organization... [the centre of] deep learning AI algorithms. Bacteria colonies, bee colonies, ant colonies... all show very advanced swarming behaviour.’  (Kasapoglu and Kurdemir 2) This does not mean that viewers need to know the details of the close relationship between robotic and biological behaviour that mark the current phase of AI. It does mean, however, that such visual sequences broadly draw the attention of the viewer well beyond the range of the plot-line to highlight the way in which cross-species cooperation is necessary if the human world --- even as represented by Ant Man, a particular kind of superhero --- is to survive.

Another sequence in the film also reaches out to take the viewer well beyond the limits of immediate cinematic narrative and action. This is an extended sequence toward the end of the film in which --- after the defeat and death of the villain Kang --- variants of the character throughout history appear and give him a chance to re-state his aims. In life and in death, Kang maintains that his sole purpose is to develop successive realms each populated selectively with a more heroic set of people, with greater capabilities and higher goals. This sounds like a nightmarish recreation of the ubermensch --- the Nietzcshean super race --- that human history knows all too well. At the same time, it also reminds the viewer that the impulse toward perfectibility of race and species --- that Kang claims is behind his lust for power --- is one that super villains share with superheroes. The same zeal for perfectibility --- no matter what the cost --- is one that influences the development of AI as well. Again, what the viewer gets in this sequence involving the historical lineage of evil is not just the instrumental virtuosity of AI but the allegorization of perfectibility. The former is responsible for the range of images, colour and animation that ‘produce’ Kang. The latter is responsible for indicating the way in which such a shared impulse may reach for the superhuman only to fall below the level of the bestial. Again, it is a sequence that reaches out and grips the viewer independent of plot and immediate necessity, so that the viewer’s gaze--- without any need for external explanation --- turns inward instead.

Such sequences lead the viewer to speculate not just on the multiple characters, plots, and narrative-groupings in the film but also on the nature of the cinematic multiverse that holds them together.  After all, this is one of the critical challenges of multiversal reality, to think about whether there is ‘one best possible world, or ... more than one unsurpassable world, or... an infinite hierarchy of infinitely better worlds.’ (Kraay 355) (9) Kang of course stakes his claim to heroism by saying that he wished to open up vistas of ever-improving worlds. Subsequently, as Lang goes to join his team at a party, he vaguely recalls Kang’s threat that --- balked of his desire for increasingly perfect worlds --- he will cause further destruction instead. Lang shrugs this off, but again it serves to remind the reader that one of the first conditions to organise the multiverse is an updated version of the old saying, corruptio optimi pessima, or, the corruption of the best becomes the worst. Under certain conditions malignancy might well serve as the other face of perfectibility. This interiorised version of intelligence headed in the wrong direction acts as a pointer to the way in which AI that produces superheroes and super villains equally, may well be an actualization of the same philosophical dilemma.

2.4 Guardians of the Galaxy Vol 3
(5) At the same time, however, the metaphysical conflict between good and evil may also serve as an index to the nature of reality that cinema can mobilize. While Ant Man and the Wasp: Quantumania relies on a combination of external and internal representations of this conflict, Guardians of the Galaxy Vol 3 takes forward the question of reality from the beginning by positioning Rocket --- the hybridized product of an experiment undertaken by the High Evolutionary on a baby racoon on whom he bestows a computerized brain --- as the current leader of the Guardians. This is a complicated situation. Rocket abuses those who identify him as a racoon. He deems it a deep insult regardless of whether friend or foe hails him by this term.  At the same time though, he bonds affectionately with all creatures –- an otter, a walrus and a rabbit --- who are hapless victims of the High Evolutionary’s efforts to develop a super race and a perfect realm within which to domicile it. When an attack short-circuits Rocket’s brain, leaving him for dead, his team has to travel to the manufacturers, Orgocorp, the company that helped create Rocket to restore him to health. The question that persists, however, concerns the nature of the life and reality that Rocket enjoys. It is not enough to describe him as a sentient being and be done with it, because that will open the door to questions not just of identity but of the nature of responsibility that is contingent on identity as well. To label Rocket as an anthropomorphic creature, an inhabitant of a posthuman world does not take the argument very far because this is at variance with the way in which he and his team respond to his role. On a limited basis we might hold on to the idea that this is an animated character placed within the framework of live action. Again though, this does not advance matters very much, because the nature of the debate on reality in Guardians of the Galaxy is nowhere near the back-and-forth dialectic between animation and the everyday that characterises say, Who Framed Roger Rabbit? 13 (Zemeckis, 1988)  It is very much more sophisticated and disturbing. Toward the end, when Rocket looks up the meaning of the term ‘racoon’ in the dictionary, it mortifies him to find that his lineage is squarely that of the animal kingdom, while he has always insisted on his status as a human being. If Rocket cannot find a lexical category that he is willing to accept, how can he or his team submit to conventional moral or ethical categories? This is not simply to point out that computer-generated imagery that stresses the instrumentality of AI brings the viewer to a world that is only just beginning to reveal itself. It is to underscore the fact that --- once the narrative disturbs the known premises of character, identity and language --- ethical categories of good and evil merit not only redefinition but perhaps retirement as well. Different kinds of reality will require different assessments.

Conclusion
This becomes clear as the film draws to its close, with the guardians dividing forces on the basis of the separate missions that they wish to undertake. When Peter Quill heads back to Earth to watch his grandfather through an episode of assisted living, the viewer may wonder what became of the debate between race versus species. The earlier option of a neat conclusion to an engagement with superheroism is no longer open. The possibilities that come together in (14)Umberto Eco’s comforting analysis of superheroism no longer apply. ‘Superman’s civic attitude is perfect, but it is exercised and structured within the sphere of a small community... as a model of absolute fidelity to establish values.’ (Eco 22)  It is not merely that Lang and Quill are known to function on the far side of the law. More significantly, the films under discussion do not so much as trivialise superheroism as use to as a means to an end. This larger purpose is not entirely clear, but goes down to the bone in redrawing the relationship between race and species.  It is true that --- rather like the closing scene in Ant Man and the Wasp: Quantumania --- there is an understated but ever-present danger in Guardians of the Galaxy: Vol 3 toward which the dialogue gestures. At the same time, it is clear that the film does not wish to push debate beyond this point, which is why the claim that these are both concluding films in their separate sequences becomes important. Is this eternal deferral of closure a signal of continuing commercial value for the franchise? Or is it because there is a stage when the anxieties of the scientific community cannot translate any further into the melodrama of the popular imagination?  The spirit of every age has its own tipping-point, beyond which it turns its face away and declines to pursue an inquiry any further. The Victorians --- satisfied with the application of a species-argument to racial difference --- did not continue indefinitely with social Darwinism. It is possible that our era can take this much exposure to the matter of species-jump, and no more. Perhaps the popular imagination can only negotiate the shift that AI makes --- from cinematic instrument to cinematic protagonist --- one instalment at a time.
----------------------------------------------------------------------------------------------------------------
Christel R Devadawson is Professor, Department of English University of Delhi. She got her PhD from Cambridge University in 1992. She is the author of, among other books and writings, Out of Line: Cartoons, Caricature and Contemporary India (Orient Blackswan, 2015)
-------------------------------------------------------------------------------------------------------

Works Cited