THE ALGORITHMIC AUTEUR

The rapid diffusion of generative artificial intelligence into the creative industries has intensified debates surrounding authorship, originality, and cultural diversity. Public discourse frequently frames these technologies either as revolutionary creative agents capable of transforming cultural production or as existential threats to human expression. Both positions, however, obscure a more consequential question: how generative AI is embedded within existing economic and institutional structures, and how this embeddedness shapes cultural production at scale.

This essay argues that the cultural effects of generative AI cannot be understood in isolation from platform capitalism - the dominant organisational form through which contemporary culture is produced, circulated, and valued. Rather than displacing human creativity outright, generative systems intensify pre-existing logics of data-driven optimisation, algorithmic visibility ranking, and engagement maximisation. Within this context emerges what is conceptualised here as the Algorithmic Auteur: not a machine author, but a distributed infrastructural system that conditions which cultural outputs are most likely to be produced, recognised, and rewarded within platform-mediated environments.

Cultural homogenisation under generative AI is therefore neither technologically inevitable nor primarily the result of biased algorithms or datasets. As used here, cultural homogenisation refers to the convergence of stylistic forms and the narrowing of what achieves visibility and reward under platform metrics. This homogenisation reflects a systemic pattern produced by the integration of generative technologies into platform systems that organise cultural production around scalability, predictability, and metric-driven valuation. By situating generative AI within the political economy of platform capitalism, this essay demonstrates how statistically dominant and culturally legible forms are systematically favoured, while also showing that these outcomes remain contingent on alternative institutional and governance arrangements.

2 Platform Capitalism and the Restructuring of Cultural Production

To understand why generative AI exhibits a tendency toward cultural homogenisation, it must be situated within the political economy of platform capitalism rather than treated as an autonomous creative technology. As Srnicek (2017) argues, platforms are not neutral intermediaries but infrastructural actors that reorganise markets around data extraction, scalability, and continuous optimisation. Their business models depend on capturing user interaction, translating it into behavioural data, and deploying that data to predict and shape future activity. When extended to cultural production, these logics transform not only how culture circulates, but also how it is conceived, produced, and evaluated from the outset.

This restructuring is captured by Poell et al.’s concept of the platformisation of cultural production, through which creative industries become structurally dependent on platform infrastructures for visibility, monetisation, and legitimacy (Poell et al., 2017). Algorithms governing recommendation, ranking, and monetisation increasingly mediate access to audiences, rendering creative success contingent upon alignment with opaque and continuously shifting platform metrics. As a result, cultural producers are incentivised to anticipate algorithmic preferences (such as format, pacing, or genre) rather than pursue aesthetic experimentation, displacing creative decision-making from autonomous judgement toward data-driven optimisation.

Nieborg and Poell (2018) further theorise this condition through the notion of the contingent cultural commodity. Under platform capitalism, cultural goods are produced as modular and provisional outputs that remain open to continuous adjustment in response to real-time feedback. Rather than stable works with fixed meanings or values, cultural products become iteratively optimised artefacts whose worth is increasingly measured through engagement indicators such as clicks, shares, watch time, or completion rates. Importantly, this transformation predates generative AI, shaping the economic and infrastructural environment into which generative systems are introduced.

Generative AI therefore intensifies rather than initiates these cultural logics. By automating the large-scale production of culturally legible outputs, generative systems operationalise the contingent logic already embedded within platform capitalism. As Berry (2025) observes, synthetic media increasingly functions through algorithmic feedback loops in which production and valuation are tightly coupled, privileging outputs that are iterable, recognisable, and optimised for platform metrics. Cultural homogenisation thus emerges as a structural tendency produced by the integration of generative technologies into platform systems organised around scalability, predictability, and metric-driven valuation.

3 From Human Auteur to Algorithmic Auteur

The emergence of generative AI does not simply replace the human auteur with a machine author; rather, it reorganises authorship into a distributed and infrastructural process shaped by algorithmic systems, platform metrics, and institutional arrangements. Under these conditions, authorship is no longer grounded primarily in individual intention or expressive originality, but is instead distributed across networks of human inputs, automated processes, and evaluative infrastructures. The concept of the Algorithmic Auteur captures this shift: not an agent that creates cultural works, but a system through which platform capitalism conditions which cultural outputs are most likely to be produced, recognised, and valued - thereby structuring tendencies toward homogenisation.

Empirical work by Jago and Carroll (2023) illustrates how algorithmic mediation destabilises conventional practices of authorship attribution. Examining disputes over credit and ownership in algorithmically assisted production, they show that authorship is not eliminated but redistributed across users, developers, platforms, and datasets. Crucially, this redistribution is formalised through legal, institutional, and commercial mechanisms that assign recognition and ownership, rendering authorship less an expressive status than an administrative designation aligned with systems of technical and economic governance.

This reconfiguration aligns with Levina and Arriaga’s (2014) account of how digital platforms restructure distinction and symbolic capital. In platform-mediated cultural fields, value judgements are progressively embedded within technical systems (such as ranking algorithms, visibility metrics, and engagement indicators) rather than derived from aesthetic evaluation or professional consensus. Distinction is thus produced through infrastructural positioning rather than artistic differentiation, privileging cultural outputs that conform to platform logics of scale, predictability, and circulation.

Nieborg and Poell’s (2018) concept of the contingent cultural commodity provides continuity between these shifts in authorship and the rise of generative AI. Cultural products are already modular and provisional, designed for ongoing revision in response to platform feedback. Generative AI intensifies this condition by automating the production, recombination, and optimisation of cultural forms at scale, further decentring human authorship. While human actors continue to direct, curate, and interpret algorithmic outputs (Palani & Ramos, 2024), their agency is increasingly exercised within constraints set by algorithmic evaluation and economic optimisation rather than through autonomous authorship alone.

4. Mechanisms of Cultural Homogenisation under the Algorithmic Auteur

4.1 Algorithmic Curation and Feedback Loops

Prior to the introduction of generative AI, algorithmic recommendation systems had already begun to narrow cultural diversity by shaping patterns of consumption and, indirectly, production. Empirical evidence from Anderson et al.’s large-scale study of Spotify demonstrates that algorithmic recommendation reduces the diversity of music consumption relative to user-led exploration, systematically favouring familiar and popular content over novelty (Anderson et al., 2020). While such systems are frequently framed as neutral tools of discovery, their design prioritises engagement and retention, producing feedback loops in which frequently consumed content becomes increasingly visible, while less popular or experimental material is progressively marginalised.

Crucially, these consumption dynamics feed back into production incentives. Born et al. (2021) show that musicians and producers actively adapt their creative choices in anticipation of platform algorithms, adjusting genre, structure, tempo, and sonic features to improve algorithmic visibility. Recommendation systems therefore do not merely reflect existing tastes; they reshape cultural production by privileging formats and styles that perform well within ranking and recommendation systems. In this recursive loop, consumption data informs recommendation, which in turn conditions production. This constitutes a core mechanism through which cultural homogenisation is reproduced under the Algorithmic Auteur.

4.2 Generative AI and Stylistic Convergence

Generative AI intensifies these homogenising dynamics by automating the reproduction of dominant cultural and stylistic patterns embedded in large-scale training data. Agarwal et al. (2025) provide empirical evidence that AI-assisted writing converges toward Western stylistic conventions, even when users attempt to maintain distinct linguistic or cultural registers. This convergence does not result from deliberate standardisation by developers, but from probabilistic modelling techniques that privilege statistically dominant patterns within training datasets.

Complementing this, Hu et al. (2025) demonstrate that large language models consistently reproduce culturally salient patterns present in their training corpora, reinforcing dominant modes of expression across diverse social and linguistic contexts. Taken together, these findings indicate that stylistic convergence emerges from statistical regularities rather than intentional design choices. Under the Algorithmic Auteur, homogenisation emerges through automated pattern replication. Cultural difference is increasingly flattened by optimisation toward the most frequent, legible, and reproducible forms encoded in data. This flattening intensifies when such outputs are subsequently rewarded within platform-mediated systems of visibility and valuation.

4.3 Optimisation, Attention, and Risk Aversion

The homogenising effects of algorithmic systems are further reinforced by platform optimisation logics that incentivise predictability and discourage creative risk. Morris (2020) shows that music platforms rely on performance metrics (such as skip rates, completion rates, and playlist inclusion) to evaluate content, encouraging artists to conform to genre norms and structural conventions associated with algorithmic success. As a result, creative experimentation becomes economically risky, while stylistic conformity is systematically rewarded.

These incentives operate within broader attention markets characterised by scarcity and intense competition. As Hwang (2020) argues, platforms externalise creative risk onto producers while internalising the value generated by user attention, generating strong incentives for safe, familiar, and easily consumable content. Duffy et al. (2019) further document how precarity in platform-mediated creative industries amplifies this dynamic, compelling cultural workers to adapt their output to algorithmic expectations in order to secure visibility and income. Under the Algorithmic Auteur, cultural convergence is therefore sustained through the interaction of optimisation metrics, attention economies, and labour insecurity, rather than through any single technological determinant.

5 Counterarguments and Limits

A common objection to claims of algorithmic homogenisation is that generative AI democratises creativity by lowering barriers to entry and expanding access to cultural production. Research on human–AI collaboration documents cases in which creators use generative tools to experiment, iterate rapidly, and participate in cultural fields previously constrained by technical skill or institutional gatekeeping (Palani & Ramos, 2024). From this perspective, generative AI appears to broaden creative participation and expand stylistic possibility rather than constrain it.

A related argument holds that algorithmic systems may enhance stylistic diversity by recombining heterogeneous cultural materials at scale. Because generative models are trained on vast and internally diverse datasets, they are often framed as engines of hybridisation capable of producing novel aesthetic forms beyond the stylistic repertoires of individual human creators (Tsao et al., 2025). Empirical studies of AI-assisted creative workflows further suggest that human agency does not disappear under automation: creators continue to guide, curate, and interpret algorithmic outputs through iterative human–machine interaction, with AI functioning primarily as a tool for exploration and early-stage ideation rather than as a substitute for authorship (Sanato & Wijayanti 2024).

These arguments identify genuine affordances of generative AI. However, their explanatory reach is limited when situated within the political economy of platform capitalism. Evidence of creative empowerment tends to emerge in experimental or non-commercial contexts at the margins of dominant platforms, where creators retain greater autonomy over evaluation and circulation (Zeilinger, 2021).

By contrast, research on platformisation shows that commercial platforms organise cultural production around engagement-driven optimisation, monetisation, and scalability. This renders cultural commodities contingent on metric-based valuation regimes, privileging outputs that are legible to platform analytics and recommendation systems (Nieborg & Poell, 2018). Alternative governance arrangements (such as commons-based digital infrastructures and cooperative platforms) demonstrate that different institutional configurations can partially decouple cultural production from engagement metrics, supporting more pluralistic outcomes by prioritising collective value and public interest over scale and profitability (Poell, Nieborg & Duffy, 2021).

The implication is not that generative AI inherently homogenises culture, nor that creative agency is extinguished under algorithmic mediation. Rather, homogenisation emerges as a structural tendency when generative technologies are deployed within platform systems that reward predictability, recognisability, and scalable engagement. The cultural consequences of generative AI therefore depend less on the technology itself than on the institutional and economic arrangements through which it is organised, evaluated, and monetised.

6 Conclusion

This essay has argued that generative AI does not inaugurate a wholly new regime of cultural production, nor does it inevitably erode creativity through technological determinism. Instead, it gives rise to what has been conceptualised as the Algorithmic Auteur: a distributed system of authorship emerging at the intersection of platform capitalism, optimisation metrics, and generative technologies. Within this configuration, creative authority is increasingly exercised through infrastructural systems that shape visibility, valuation, and stylistic legibility, rather than through individual artistic intention alone.

Cultural homogenisation under these conditions is best understood as a structural tendency rather than a technological fate. Recommendation systems, attention markets, and data-driven optimisation long predate generative AI, but the automation and scale introduced by generative systems intensify their effects, systematically privileging recognisable and statistically dominant cultural forms. These outcomes do not arise from malign algorithmic intent or the disappearance of human agency, but from economic and institutional arrangements that reward predictability, engagement, and scalability over experimentation and divergence.

This distinction has significant implications. As critical political economy perspectives emphasise (Fuchs, 2014), technological systems reflect and reproduce the social relations in which they are embedded. Debates about generative AI’s cultural impact therefore cannot be resolved through technical fixes or model-level interventions alone. The future of cultural diversity in the age of generative AI will depend less on the capabilities of the technology itself than on the institutional choices through which culture is organised, evaluated, and rendered economically viable under platform capitalism.

Bibliography

Agarwal, D., Naaman, M., & Vashistha, A. (2025, April). AI suggestions homogenize writing toward western styles and diminish cultural nuances. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems (pp. 1-21).

Anderson, A., Maystre, L., Anderson, I., Mehrotra, R., Lalmas, M. and Van Der Schaar, M. (2020). Algorithmic effects on the diversity of consumption on Spotify. Proceedings of the Web Conference 2020. New York: ACM, pp. 2155–2165.

Berry, D.M. (2025). Synthetic media and computational capitalism: Towards a critical theory of artificial intelligence. AI & Society, 40(5), pp. 5257–5269.

Born, G., Morris, J., Diaz, F., & Anderson, A. (2021). Artificial intelligence, music recommendation, and the curation of culture.

Duffy, B.E., Poell, T. and Nieborg, D.B. (2019). Platform practices in the cultural industries: Creativity, labor, and citizenship. Social Media + Society, 5(4), pp. 1–13.

Fuchs, C. (2014). Digital Labour and Karl Marx. London: Routledge.

Hwang, T. (2020). Subprime attention crisis: Advertising and the time bomb at the heart of the Internet. FSG originals.Hu, T., Kyrychenko, Y., Rathje, S. (2025). Generative language models exhibit social identity biases. Nat Comput Sci 5, 65–75.

Jago, Arthur S., and Glenn R. Carroll. “Who made this? Algorithms and authorship credit.” Personality and Social Psychology Bulletin 50.5 (2024): 793-806.

Levina, N., & Arriaga, M. (2014). Distinction and status production on user-generated content platforms: Using Bourdieu’s theory of cultural production to understand social dynamics in online fields. Information systems research, 25(3), 468-488.Morris, J.W. (2020). Music platforms and the optimization of culture. Social Media + Society, 6(3), pp. 1–12.

Nieborg, D. B., & Poell, T. (2018). The platformization of cultural production: Theorizing the contingent cultural commodity. New media & society, 20(11), 4275-4292.

Poell, T., Nieborg, D. B., & Duffy, B. E. (2021). Platforms and cultural production. John Wiley & Sons.

Poell, T., Nieborg, D., & Van Dijck, J. P. (2019). Internet Policy Review. Journal On Internet Regulation, 8.

Tsao, J., Liang, C. X., Nogues, C., & Wong, A. (2025). Perceptions and integration of generative artificial intelligence in creative practices and industries: a scoping review and conceptual model. AI & SOCIETY, 1-20.

Santoso, B., & Wijayanti, R. (2024). Human-AI Collaboration in Creative Industries: Workflows in Media Production and Community-Driven Platforms. Transactions on Artificial Intelligence, Machine Learning, and Cognitive Systems, 9(11), 11-26.

Srnicek, N. (2017). Platform capitalism. John Wiley & Sons.

Zeilinger, M. (2021). Tactical entanglements: AI art, creative agency, and the limits of intellectual property (p. 186). meson press.