Framing the Future: The Impact of AI-Generated Video on Major Film Creators

Sep 6, 2025
16 min read

Updated: 2 days ago

Author: Hannah White

Affiliation: VBNN Smart Education Group

Received 25 June 2025; Revised 10 August 2025; Accepted 20 August 2025; Available online 6 September 2025; Version of Record 6 September 2025; Update 2 May 2026.

Volume 2, December 2025, (10014 - 2)

https://doi.org/10.65326/u7y566816

Abstract

AI-generated video has moved quickly from experiment to a working component of media production. In large-scale filmmaking it promises faster ideation, lower production costs, and new modes of world-building, while raising unresolved questions about authorship, labour, cultural standardisation, and the global distribution of creative power. This article develops a multi-level conceptual account of these changes by bringing together three sociological traditions: Bourdieu’s theory of fields and capital, world-systems analysis, and the institutional theory of organisational isomorphism. The central argument is that generative capability functions as a distinct and convertible form of capital, provisionally termed algorithmic capital, which compounds the positional advantage of organisations already rich in data, compute, and distribution access, even as it lowers some barriers for smaller creators. The framework yields a set of propositions concerning capital conversion, the decoupling of production diffusion from value capture, convergence in adoption practices, the aesthetic bias embedded in training data, and the strain that distributed authorship places on existing credit and rights regimes. The analysis is interpretive rather than empirical; its contribution is to integrate micro-level field dynamics, macro-level global value chains, and meso-level institutional pressures within a single explanatory frame, and to specify indicators through which the resulting propositions can be examined empirically.

Keywords: AI-generated video; field theory; algorithmic capital; world-systems analysis; institutional isomorphism; creative labour; authorship; cultural production.

1. Introduction

The film industry sits at a structural turning point created by the convergence of artificial intelligence, virtual production, and platform distribution. Systems that synthesise moving images, voices, and visual styles from high-level instructions now touch every stage of production: storyboarding and concept art in pre-production; virtual sets and synthetic doubles during the shoot; editing, clean-up, and localisation in post-production; and the rapid generation of marketing variants for distribution (Russell & Norvig, 2021; Anantrasirichai & Bull, 2022). For large studios, these tools promise scale, speed, and cost reduction. For smaller creators, they can open forms of cinematic expression previously foreclosed by budget.

The same properties that make generative systems attractive also unsettle long-standing norms. When a convincing scene can be produced from a text prompt, the boundary of artistic authorship becomes harder to locate. When synthetic doubles substitute for performers, established labour protections come under pressure. And because models learn from large corpora of films, images, and recorded performances, questions of permission, compensation, and attribution become unavoidable (Crawford, 2021; Floridi, 2023).

Research gap. Scholarship on AI in the creative industries has developed along two largely separate tracks. The first is technical, mapping the capabilities of generative models across the production pipeline (Anantrasirichai & Bull, 2022). The second is normative, concerned with bias, copyright, consent, and environmental cost (Bender et al., 2021; Crawford, 2021; Floridi, 2023). Sociological theories of cultural production, in turn, have tended to treat digital tools as instruments operating within a field rather than as resources that restructure the field itself (Bourdieu, 1993; Hesmondhalgh, 2019), while political-economy and world-systems accounts of media rarely engage the specific mechanics of generative production pipelines (Mosco, 2009; Wallerstein, 2004). What is missing is an integrated framework that connects three questions usually examined in isolation: how generative video reorganises competitive positions among film-industry agents; how it redistributes creative labour and value across the global economy; and why organisations converge on similar adoption practices under conditions of uncertainty. This article addresses that gap.

Aim and contribution. The aim is to construct a multi-level conceptual framework for the impact of AI-generated video on major film production and to translate it into propositions amenable to empirical scrutiny. The contribution is threefold. First, the article introduces the concept of algorithmic capital and locates it within Bourdieu’s scheme of capital conversion. Second, it links field-level competition to the global division of creative labour through world-systems analysis, and to convergent organisational behaviour through institutional theory, thereby joining the micro, macro, and meso levels of analysis. Third, it derives a set of propositions and corresponding indicators intended to guide subsequent empirical research on generative cinema.

2. From Digital to Generative Cinema

Digital technologies have shaped cinema for decades. Non-linear editing, computer-generated imagery, motion capture, and LED volumes have steadily reduced physical constraints and expanded what can be represented on screen (Manovich, 2001). The current shift is qualitatively different. Earlier digital tools enhanced footage that human crews had captured; generative models instead produce moving images, voices, and styles directly from high-level prompts. The salient change is generativity rather than mere manipulation.

Three features matter for large-scale filmmaking. The first is elastic scale: once trained, a model can generate many alternatives at low marginal cost, allowing rapid iteration of story beats, camera angles, and tone. The second is stylistic control: with reference conditioning and prompt engineering, teams can hold a consistent visual language across sequences. The third is localisation at volume: dialogue replacement, accent adaptation, and culturally tuned set dressing can be produced for global releases with comparatively little additional labour.

These features alter not only craft but also the political economy of cinema. They shift bargaining power, reconfigure supply chains, and unsettle the meaning of originality. Walter Benjamin’s account of how mechanical reproduction erodes the aura of the unique work anticipates the central tension of generative cinema, where the capacity to reproduce and recombine style at scale strains the symbolic value attached to singular human craft (Benjamin, 2008; Manovich, 2020). The analytical task is therefore to treat generative video not as an isolated tool but as a force that reorganises positions, value, and norms across the industry.

3. Conceptual Framework and Method

3.1 Research design

This is a conceptual, theory-building study. Rather than test hypotheses against new data, it integrates established sociological frameworks to explain an emerging phenomenon and to generate propositions for later empirical work. The approach is appropriate when a phenomenon is recent, fast-moving, and not yet well captured by existing measures, and when the contribution sought is explanatory structure rather than descriptive estimation. The analysis proceeds by interpretation and argument; its claims are conditional and are presented as such.

3.2 Selection of frameworks

Three frameworks were selected because they operate at complementary levels of analysis and, taken together, cover the questions identified in the research gap. Bourdieu’s field theory addresses the micro and meso dynamics of competition for position and the conversion of different forms of capital (Bourdieu, 1984, 1993). World-systems analysis addresses the macro structure of a global division of labour between core, semi-periphery, and periphery (Wallerstein, 2004). The institutional theory of isomorphism addresses why organisations facing shared uncertainty come to resemble one another in their practices (DiMaggio & Powell, 1983; Scott, 2014). Frameworks confined to a single level, or to the technical or purely ethical dimensions of AI, were judged insufficient because the phenomenon spans competition, global structure, and organisational behaviour simultaneously.

3.3 Analytical procedure

The procedure has three steps. First, the features of generative video identified in Section 2 are mapped onto the mechanism each framework specifies, producing the correspondences summarised in Table 1. Second, from each mapping a small number of propositions is derived, stated so that they could in principle be confirmed or disconfirmed by evidence. Third, the propositions are examined together to identify points of tension and reinforcement across levels, which informs the discussion of theoretical contribution. Illustrative scenarios are used as heuristic devices to clarify mechanisms; they are analytic constructions, not empirical case studies, and no quantitative claims are attached to them.

3.4 Scope and boundary conditions

The framework is bounded to large-scale, commercially financed film production and its immediate periphery of vendors, platforms, and independent creators competing for the same audiences and festivals. It does not attempt to cover all moving-image media, nor does it forecast the trajectory of particular technologies. The propositions concern structural tendencies under current conditions of uneven access to data, compute, and distribution; they would weaken if those conditions changed, for example through broad availability of capable open models or binding international rules on training data.

Table 1. Analytical mapping of generative video onto the three frameworks

Framework	Level of analysis	Core mechanism	Expected dynamic in generative cinema
Field theory (Bourdieu)	Micro / meso	Competition for position; conversion among economic, cultural, social, and symbolic capital	Generative capability acts as a new convertible resource that compounds incumbents’ advantage and reshapes claims to prestige
World-systems analysis (Wallerstein)	Macro	Global division of labour and unequal exchange between core, semi-periphery, and periphery	Production capacity diffuses outward while value capture and standard-setting remain concentrated in the core
Institutional isomorphism (DiMaggio & Powell; Scott)	Meso	Coercive, mimetic, and normative pressures toward convergence	Studios and vendors adopt similar AI practices, credentials, and clauses even where strategic value is unproven

Note. The table summarises the correspondences developed in Sections 4–7. Levels of analysis are indicative rather than mutually exclusive; the three frameworks overlap, and the discussion treats their intersections explicitly.

4. Algorithmic Capital and the Field of Film Production

Bourdieu describes cultural production as a field in which agents compete for position using economic, cultural, social, and symbolic capital (Bourdieu, 1993). In blockbuster filmmaking, major studios and platforms typically command abundant economic capital through financing, strong social capital through distribution relationships, and high symbolic capital through brand prestige, while independent creators often rely on cultural capital, expressed as distinctive taste or an experimental ethos, to gain recognition (Bourdieu, 1984; Hesmondhalgh, 2019).

Generative systems introduce a hybrid resource that is not reducible to the existing four. I term it algorithmic capital: the accumulated technical assets that enable superior generative outcomes, including proprietary or licensed datasets, compute infrastructure, fine-tuned models, and the organisational know-how to integrate them into production pipelines. Algorithmic capital matters analytically because it is convertible. It lowers production costs and so converts into economic capital; it enables distinctive looks and workflows and so converts into cultural capital; it attracts collaborators and so converts into social capital; and it can yield awards or critical attention and so converts into symbolic capital. Its accumulation depends on prior endowments of data, compute, and distribution access, which means that those already advantaged in the field are best placed to acquire it.

Because the assets that constitute algorithmic capital are the same assets that platform firms accumulate as a matter of business model, holders of large data and compute reserves can compound their positions (Srnicek, 2017; Zuboff, 2019). Studios with extensive intellectual-property libraries can generate high-fidelity variations that remain consistent with an established brand, reinforcing symbolic capital. Platforms with detailed audience data can anticipate responses and convert that knowledge into economic returns. Vendors who master provenance tagging and rights clearance accrue normative legitimacy as responsible practitioners, a further source of symbolic capital and an emerging form of gatekeeping (Gillespie, 2018). The first proposition follows.

P1. Generative capability functions as a distinct and convertible form of capital whose accumulation compounds the positional advantage of organisations already rich in data, compute, and distribution access.

The same framework clarifies a countervailing dynamic around authenticity. Audiences attach symbolic value to perceived craft, risk, and embodied performance, the very qualities that Benjamin associated with aura (Benjamin, 2008). Where generative techniques are concealed and the result is read as mechanically smooth but emotionally thin, symbolic capital can erode. Where, by contrast, human decision-making is made visible through disclosed craft notes, annotated credits, and behind-the-scenes material, a form of hybrid authorship can preserve or even enhance prestige. Restraint and transparency thus become resources in the competition for symbolic capital.

P2. Where AI use is disclosed and human authorship is foregrounded, symbolic capital is preserved or enhanced; where it is concealed, perceived authenticity and the associated symbolic capital tend to erode.

5. Core Consolidation and Semi-Peripheral Ascent

World-systems analysis distinguishes a core that controls high-value activity and advanced technology, a semi-periphery with mixed capabilities, and a periphery that supplies resources and lower-cost labour (Wallerstein, 2004). In cinema, the core has historically retained high-margin intellectual property, marketing, and global distribution, while peripheral regions have supplied lower-cost tasks such as rotoscoping and asset clean-up, as well as locations (Mosco, 2009).

Generative tools move this map in two directions at once. On one side, production capability diffuses outward: semi-peripheral creators can now generate scenes that once required expensive equipment, enabling competitive entries at festivals and on streaming services. On the other side, value capture consolidates: the frontier models, the compute on which they run, the data licensing deals, and the distribution platforms remain largely core-controlled. Even when the act of production disperses, the rent-bearing layers of model hosting, promotion, and monetisation tend to stay in the core. Production diffusion and value diffusion therefore come apart.

P3. The diffusion of generative production capacity to semi-peripheral creators coincides with the consolidation of value capture and standard-setting in core-controlled model and distribution layers, decoupling production diffusion from value diffusion.

Data flows add a cultural dimension to this unequal exchange. When models are trained predominantly on cultural products originating in the core, their stylistic defaults tend to privilege core aesthetics. Peripheral creators gain access to powerful tools but risk reproducing dominant visual grammars, a form of cultural dependency rather than autonomy (Crawford, 2021; Bender et al., 2021). A fairer arrangement would require consent-based and compensated training data drawn from a wider range of traditions, together with governance that allows local styles to inform model priors.

P4. Models trained predominantly on core cultural corpora encode aesthetic priors that bias generative defaults toward dominant styles, reproducing cultural dependency in the absence of corrective data governance.

Finally, the material substrate of generative cinema is unevenly distributed. Compute-intensive training and rendering concentrate advantages in firms with capital and platform arrangements, producing a compute gap that maps onto existing inequalities, while the environmental burdens of energy use and hardware supply chains often fall on semi-peripheral and peripheral regions (Crawford, 2021). The structural point is that access to compute is itself a stratifying resource, reinforcing the dynamics described in P3 and P4.

6. Institutional Convergence in AI Adoption

Why do organisations facing the same technology come to adopt strikingly similar practices? Institutional theory identifies three pressures toward convergence (DiMaggio & Powell, 1983; Scott, 2014). Coercive pressure arises from law, contracts, and the requirements of powerful counterparties: studios adopt watermarking, provenance tags, and consent clauses because insurers, distributors, and regulators increasingly demand them, and once a few dominant buyers make such documentation a precondition for deals, others conform. Mimetic pressure arises from uncertainty: when a prominent production uses generative techniques for de-aging or multilingual dubbing with apparent success, competitors imitate the practice to reduce perceived risk and to signal modernity, and the playbooks, checklists, and budget lines that encode the practice spread quickly. Normative pressure arises from professional education and standards bodies that socialise practitioners into shared expectations about disclosure, ethics review, and crediting; over time, AI literacy becomes a credential, and those who lack it may be excluded from prestige projects.

The implication is that adoption can become widespread for reasons that are partly independent of demonstrated strategic value. Convergence is driven as much by legitimacy and risk management as by efficiency, which helps explain why similar workflows appear across firms with very different resources (Floridi, 2023; UNESCO, 2021).

P5. Under conditions of uncertainty, studios and vendors converge on similar AI practices, credentials, and contractual clauses through coercive, mimetic, and normative pressures, producing institutional isomorphism in adoption even where strategic value remains unproven.

7. Creative Labour, Authorship, and Provenance

The three structural frameworks meet in the experience of creative work. Generative pipelines do not eliminate creative labour uniformly; they redistribute it. Some routine tasks, such as rotoscoping and wire removal, are increasingly automated. Others are augmented rather than replaced: editors act as coordinators of generative outputs, visual-effects artists supervise model behaviour, and production designers curate synthetic assets. New specialised roles emerge alongside the traditional ones, including prompt design, dataset curation, and the coordination of rights and provenance (Anantrasirichai & Bull, 2022; Hesmondhalgh & Baker, 2011).

At the same time, generative workflows extend managerial visibility into creative work. Time-stamped versioning and productivity dashboards can standardise craft into measurable units, creating a risk of de-skilling and of piece-rate pressure on individual contributors (Pasquale, 2020). The outcome is not predetermined. Well-designed pipelines can also elevate human judgement by relieving artists of repetitive work and rewarding discernment, and collective agreements can set pay floors for synthetic stand-ins, define reuse windows for digital doubles, and require credit for the data labour of performers and artists whose contributions inform a model (Hesmondhalgh & Baker, 2011).

P6. Generative pipelines simultaneously de-skill routine creative tasks and create new specialised roles, shifting rather than uniformly eliminating creative labour, with the balance between de-skilling and up-skilling shaped by pipeline design and collective bargaining.

Authorship is reconfigured in parallel. Generative cinema is inherently distributed: model designers, data contributors, prompt authors, editors, and performers all shape the result. In place of a single author there is a layered authorship that strains credit and rights regimes built around the individual creator (Lessig, 2004; Boyle, 2008; Benkler, 2006). Ethical pipelines require verifiable consent from performers for likeness and voice, from artists for stylistic reference, and from rights holders for adjacent intellectual property, ideally through opt-in datasets with tiered licensing. Technical provenance, whether through metadata, cryptographic signatures, or watermarking, allows asset histories to be traced, supporting legal compliance, audience trust, and scholarly study, and aligning with emerging participatory and co-creative relationships between studios and audiences (Jenkins, 2006).

P7. Layered authorship in generative cinema strains single-author credit and rights regimes, generating pressure toward proportional, provenance-based attribution and compensation.

8. Discussion

The framework makes several contributions to the theories it draws upon and to current debate. For field theory, the concept of algorithmic capital extends Bourdieu’s account beyond the four classical forms of capital to a technical resource that is both convertible and dependent on prior endowments. This matters because it specifies a mechanism by which a new technology reproduces, rather than disrupts, existing hierarchies in cultural production: those best able to accumulate algorithmic capital are those already advantaged, and the convertibility of that capital lets advantage compound (Bourdieu, 1993; Hesmondhalgh, 2019). The treatment of disclosure and restraint as resources in the competition for symbolic capital also refines the classic argument about aura, suggesting that the value of perceived authenticity rises rather than falls as reproduction becomes effortless (Benjamin, 2008).

For world-systems analysis, the framework identifies a specific contemporary form of unequal exchange in which production capability and value capture are decoupled. The contribution is to show that the apparent democratisation of production need not redistribute economic or symbolic returns, because the rent-bearing layers remain core-controlled and because training data encode core aesthetics (Wallerstein, 2004; Mosco, 2009; Crawford, 2021). This sharpens debates about whether generative tools empower peripheral creators by distinguishing access to tools from control over value.

For institutional theory, the framework applies the classic typology of isomorphic pressures to a setting of acute technological uncertainty and shows how legitimacy and risk management, rather than demonstrated efficiency, can drive the rapid spread of practices across heterogeneous firms (DiMaggio & Powell, 1983; Scott, 2014). The wider implication is that governance norms, once embedded in contracts and professional standards, may shape the industry’s trajectory at least as strongly as the underlying capabilities.

Read together, the propositions speak directly to the debate over whether AI will democratise or concentrate creative power. The answer suggested here is that it does both, but along different axes: it diffuses the ability to make images while concentrating the ability to capture value and set standards. A governance agenda follows from the analysis rather than being appended to it. Foregrounding human decision-making preserves symbolic capital (P2); consent-based and compensated data governance addresses the cultural dependency identified in P4; transparent, provenance-based crediting responds to the strain on authorship in P7; and labour protections mediate the de-skilling risk in P6. These are not neutral technical fixes but interventions in the distribution of capital, value, and recognition that the framework describes (Floridi, 2023; UNESCO, 2021).

9. Limitations and Future Research

This study is conceptual and interpretive, and its claims should be read accordingly. The propositions are arguments derived from theory, not findings established against data, and the illustrative scenarios are heuristic constructions rather than evidence. The three frameworks, though complementary, originate largely in Western social theory and may underweight aesthetic and institutional dynamics elsewhere. Because the technology is changing quickly, the boundary conditions stated in Section 3.4 are consequential: broad availability of capable open models or binding rules on training data would weaken several propositions.

These limitations indicate a clear empirical agenda. The propositions can be operationalised and tested through ethnographies of visual-effects houses and editorial teams, analysis of production contracts and guild agreements, and audits of training datasets and model outputs. Several indicators would support such work: the share of creative and economic value accruing to semi-peripheral collaborators (relevant to P3); the degree to which generative defaults reproduce dominant styles, measured through systematic output audits (P4); the proportion of automated tasks paired with funded up-skilling and stable wages (P6); the share of final assets carrying complete, machine-readable provenance (P7); and survey measures of audience-perceived transparency and authenticity (P2). Comparative work across national film industries would test the generality of the framework beyond its current scope, and longitudinal study would show whether the convergence predicted in P5 stabilises or gives way to renewed differentiation.

10. Conclusion

AI-generated video does not mark the end of cinema but the beginning of a new phase, in which the means of producing spectacle and story change while their importance does not. Read through field theory, world-systems analysis, and institutional theory, generative video reorganises large-scale filmmaking along a new axis of advantage, algorithmic capital, while encouraging convergence in practice. Production capability spreads outward even as value capture, standard-setting, and the aesthetic priors embedded in models remain concentrated in the core. The contribution of this article is to integrate these micro, macro, and meso dynamics in a single framework, to introduce algorithmic capital as a convertible resource within Bourdieu’s scheme, and to state propositions and indicators that make the framework empirically tractable. Whether the next era of cinema becomes more inventive and more inclusive, or simply more concentrated, will depend less on the capabilities of the models than on how the field distributes capital, value, and recognition among the people whose judgement still animates the moving image.

References

Anantrasirichai, N., & Bull, D. (2022). Artificial intelligence in the creative industries: A review. Artificial Intelligence Review, 55(1), 589–656. https://doi.org/10.1007/s10462-021-10039-7
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (pp. 610–623). Association for Computing Machinery. https://doi.org/10.1145/3442188.3445922
Benjamin, W. (2008). The work of art in the age of mechanical reproduction (J. A. Underwood, Trans.). Penguin. (Original work published 1936)
Benkler, Y. (2006). The wealth of networks: How social production transforms markets and freedom. Yale University Press.
Bourdieu, P. (1984). Distinction: A social critique of the judgement of taste (R. Nice, Trans.). Harvard University Press.
Bourdieu, P. (1993). The field of cultural production: Essays on art and literature. Columbia University Press.
Boyle, J. (2008). The public domain: Enclosing the commons of the mind. Yale University Press.
Crawford, K. (2021). Atlas of AI: Power, politics, and the planetary costs of artificial intelligence. Yale University Press.
DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48(2), 147–160. https://doi.org/10.2307/2095101
Floridi, L. (2023). The ethics of artificial intelligence: Principles, challenges, and opportunities. Oxford University Press. https://doi.org/10.1093/oso/9780198883098.001.0001
Gillespie, T. (2018). Custodians of the Internet: Platforms, content moderation, and the hidden decisions that shape social media. Yale University Press.
Hesmondhalgh, D. (2019). The cultural industries (4th ed.). Sage.
Hesmondhalgh, D., & Baker, S. (2011). Creative labour: Media work in three cultural industries. Routledge. https://doi.org/10.4324/9780203855881
Jenkins, H. (2006). Convergence culture: Where old and new media collide. New York University Press.
Lessig, L. (2004). Free culture: How big media uses technology and the law to lock down culture and control creativity. Penguin Press.
Manovich, L. (2001). The language of new media. MIT Press.
Manovich, L. (2020). Cultural analytics. MIT Press. https://doi.org/10.7551/mitpress/11214.001.0001
Mosco, V. (2009). The political economy of communication (2nd ed.). Sage.
Pasquale, F. (2020). New laws of robotics: Defending human expertise in the age of AI. Harvard University Press. https://doi.org/10.4159/9780674250062
Russell, S., & Norvig, P. (2021). Artificial intelligence: A modern approach (4th ed.). Pearson.
Scott, W. R. (2014). Institutions and organizations: Ideas, interests, and identities (4th ed.). Sage.
Srnicek, N. (2017). Platform capitalism. Polity Press.
UNESCO. (2021). Recommendation on the ethics of artificial intelligence. United Nations Educational, Scientific and Cultural Organization.
Wallerstein, I. (2004). World-systems analysis: An introduction. Duke University Press. https://doi.org/10.1215/9780822399018
Zuboff, S. (2019). The age of surveillance capitalism: The fight for a human future at the new frontier of power. PublicAffairs.