Artificial intelligence as political antagonism: media traces of a displaced conflict

Abstract Contemporary AI is insistently declared controversial, yet public debate around it rarely assumes the form of a technoscientific controversy in the sense this term carries in science and technology studies (STS)—no sustained expert dispute, no emergent lay expertise, no stable issue publics. This article argues that the diagnosis of absence mistakes the form of conflict for its existence. Conflict over AI has been displaced, from the inaccessible technical object onto the imaginaries surrounding it, and from horizontal controversy among citizens onto a vertical antagonism opposing a “people” to the institutional order AI is taken to embody. We first show that controversy analysis rests on two assumptions—an epistemic one, tying legitimate participation to expertise, and a materialist one, locating the reality of conflict in technical objects rather than symbolic formations—which render such displaced conflict illegible. Drawing on Laclau and Mouffe’s theory of hegemony, we then characterize its structure as populist in a precise, non-pejorative sense. Empirically, we analyze 1,148,092 comments posted on 4244 videos of TF1.info, the YouTube channel of France’s largest private broadcaster, comparing the publics of AI with those of electric vehicles, nuclear technologies, and climate change, and mapping their intersections with party-political publics. Quantitative mapping shows weak issue-public formation around AI; qualitative analysis reveals a vertical antagonism in which AI condenses institutional power. Recognizing this conflict, rather than eliciting the controversies “we need”, is the precondition of a democratic politics of AI. Similar content being viewed by others Data availability The data that support the findings of this study are available from the corresponding author upon reasonable request. Notes Many of the articles of this special issue emerged out from the research project “Shaping 21st Century AI: Controversies and Closure in Media, Policy, and Research.” The project was designed “to investigate how AI as a socio-technical phenomenon was taken up and imposed on contemporary societies during the formative decade from 2012 to 2021 and the role of controversies in shaping this process, in four countries: Germany, France, the United Kingdom and Canada” (Marres et al. 2025, p. 6). Our characterization of this literature as circumscribed rests on a two-pronged search conducted on Google Scholar (accessed July 1st, 2026). We first combined AI-related terms with the lexicon of controversy—(“artificial intelligence” OR “generative AI” OR “large language models”) AND (“controversy” OR “controversies”)—and found that, across the most cited and most recent results, “controversy” is overwhelmingly used in a generic sense (public disagreement, scandal) rather than in the technical STS sense of a method and object of analysis; a narrower query anchored to that technical vocabulary —(“artificial intelligence” OR “machine learning”) AND (“controversy mapping” OR “controversy analysis” OR “mapping controversies”)—returned a markedly smaller set, largely coextensive with the Big Data & Society collection cited above. We complemented this with a citation-chaining exercise on that collection, which confirmed that only a minority of citing works engage AI as a controversy in the analytical sense. One might be tempted to extend this analysis to AI itself, whose “thingness” is, as Suchman (2023) shows, a precarious and continuously renewed objectification. On such a reading, the proliferation of discursive work required to stabilize “AI” as a singular object would itself index the antagonism that traverses it. We do not pursue this ontological extension here, but it indicates the depth at which the two vocabularies—the STS critique of AI’s reification and the theory of hegemony—can communicate. The concept is specifically a Mouffean development. Elaborated in her single-authored work from The Democratic Paradox (2000) onward—see, in particular, Mouffe (2013)—, it addresses a question largely absent from Laclau’s writings—how the antagonism constitutive of the political can be given democratically sustainable form. Not all videos produced by TF1 for its television broadcast are uploaded to the TF1.info YouTube channel. In addition, platform and channel-level moderation mean that a small share of comments does not appear in the dataset. Based on observable discrepancies between comment counts and retrieved data, we estimate this proportion to be approximately 1%. However, the exact content and distribution of moderated comments remain inaccessible. The three categories were developed inductively. One of us watched all 23 videos and drafted the typology on the basis of recurrent thematic emphases in titles, descriptions, and content; the classification was then applied and reviewed jointly, with borderline cases resolved by consensus. Since several videos could plausibly fit more than one category, each was assigned to the category corresponding to its dominant emphasis. For instance: “Une voiture électrique produite avec sa batterie en Norvège […] ne devra parcourir que 8 000 km pour compenser les émissions de CO2 […] En France, avec 78% de l’énergie électrique produite par le nucléaire, le chiffre tombe à 16 800 km” [“An electric car produced with its battery in Norway […] would only need to travel 8,000 km to offset CO2 emissions […] In France, with 78% of electricity from nuclear power, the figure drops to 16,800 km”]; “Que l’électrolite soit liquide ou solide, cela n’empêche pas les contraintes thermiques […] La densité énergétique des batteries restera toujours très inférieure à l’essence […] L’électrique est antérieur au thermique, il a disparu une fois pour des raisons purement techniques et ça sera amené à se répéter” [“Whether the electrolyte is liquid or solid does not prevent thermal stress […] Battery energy density will always remain far below that of gasoline […] Electric vehicles predate combustion ones, and they disappeared once for purely technical reasons, which may well happen again”]. The qualitative analysis proceeded by close reading rather than formal coding. We read the comments posted on the 23 AI-related videos in their entirety, attending to three dimensions: (1) the presence or absence of a technical register (specialized vocabulary, references to documentation, procedural or design-level claims); (2) the target of critique (the technology itself, other commenters, or institutions, elites, and media); (3) and the discursive linking of AI to grievances external to the technology (employment, cost of living, political representation). Although no formal coding scheme was applied, these dimensions provided consistent criteria for interpretation. In particular, we treat a comment as expressing vertical antagonism—rather than ordinary negativity, irony, or diffuse distrust—when it combines the second and third dimensions, that is, when critique is directed upward at institutional actors and AI is articulated as one element within a broader chain of grievances. Negativity aimed at the technology’s performance, humor without an institutional target, and disagreement among commenters were not treated as antagonistic in this sense. The excerpts quoted below were selected as representative of patterns recurring across the corpus, not as isolated instances. References Bareis J, Katzenbach C (2022) Talking AI into being: the narratives and imaginaries of national AI strategies and their performative politics. Sci Technol Hum Val 47(5):855–881. https://doi.org/10.1177/01622439211030007 Barry A (2012) Political situations: knowledge controversies in transnational governance. Crit Policy Stud 6(3):324–336. https://doi.org/10.1080/19460171.2012.699234 Collins HM, Evans R (2002) The third wave of science studies: studies of expertise and experience. Soc Stud Sci 32(2):235–296. https://doi.org/10.1177/0306312702032002003 Dandurand G, McKelvey F, Roberge J (2023) Freezing out: legacy media’s shaping of AI as a cold controversy. Big Data Soc. https://doi.org/10.1177/20539517231219242 Epstein S (1995) The construction of lay expertise: AIDS activism and the forging of credibility in the reform of clinical trials. Sci Technol Hum Val 20(4):408–437. https://doi.org/10.1177/016224399502000402 Gobbo B, Marres N, Poletti C, Terenghi G (2025) The shape-shifter: a social design method for eliciting AI & society controversies. OSF Preprints, April 4. https://doi.org/10.31235/osf.io/ugzq8_v1 González-Arias C, López-García X (2024) Rethinking the relation between media and their audience: the discursive construction of the risk of artificial intelligence in the press of Belgium, France, Portugal, and Spain. J Media 5(3):1023–1037. https://doi.org/10.3390/journalmedia5030065 Gourlet P, Ricci D, Crépel M (2024) Reclaiming artificial intelligence accounts: a plea for a participatory turn in artificial intelligence inquiries. Big Data Soc. https://doi.org/10.1177/20539517241248093 Hilgartner S (2015) Capturing the imaginary: vanguards, visions and the synthetic biology revolution. In: Hilgartner S, Miller CA, Hagendijk R (eds) Science and democracy: making knowledge between politics and policy. Routledge, London, pp 33–55 Jasanoff S (2015) Future imperfect: science, technology, and the imaginations of modernity. In: Jasanoff S, Kim S-H (eds) Dreamscapes of modernity: sociotechnical imaginaries and the fabrication of power. University of Chicago Press, Chicago, pp 1–33 Laclau E (2005) On populist reason. Verso, London Laclau E, Mouffe C (2001) Hegemony and socialist strategy: towards a radical democratic politics, 2nd ed. Verso, London Latour B (1987) Science in action: how to follow scientists and engineers through society. Harvard University Press, Cambridge Liebig L, Jobin A, Güttel L, Katzenbach C (2024) Situating AI policy: controversies covered and the normalisation of AI. Big Data Soc. https://doi.org/10.1177/20539517241299725 Mao Y, Richter V, Katzenbach C (2025) Strategising imaginaries: how corporate actors in China, Germany and the US shape AI governance. Big Data Soc. https://doi.org/10.1177/20539517251400727 Marres N, Castelle M, Gobbo B, Poletti C, Tripp J (2024) AI as super-controversy: eliciting AI and society controversies with an extended expert community in the UK. Big Data Soc. https://doi.org/10.1177/20539517241255103 Marres N, Katzenbach C, Munk AK, Jobin A (2025) On the controversiality of AI: the controversy is not the situation. Big Data Soc. https://doi.org/10.1177/20539517251383870 Mouffe C (2000) The democratic paradox. Verso, London Mouffe C (2013) Agonistics: thinking the world politically. Verso, London Mouffe C (2018) For a left populism. Verso, London Munk AK, Jacomy M, Ficozzi M, Jensen TE (2024) Beyond artificial intelligence controversies: what are algorithms doing in the scientific literature? Big Data Soc. https://doi.org/10.1177/20539517241255107 Popa EO, Blok V, Wesselink R (2021) An agonistic approach to technological conflict. Philos Technol 34(4):717–737. https://doi.org/10.1007/s13347-020-00430-7 Rancière J (1999) Disagreement: politics and philosophy. Translated by Julie Rose. University of Minnesota Press, Minneapolis Romele A (2022) Images of artificial intelligence: a blind spot in AI ethics. Philos Technol 35(4):1–19. https://doi.org/10.1007/s13347-022-00498-3 Romele A, Severo M (2023) Microstock images of artificial intelligence: How AI creates its own conditions of possibility. Convergence 29(5):1226–1242. https://doi.org/10.1177/13548565231199982 Romele A, Severo M (2025) How effective are depictions of AI? Reflections from an experimental study in science communication. AI Soc 40(7):5127–5138. https://doi.org/10.1007/s00146-025-02283-0 Sloane M (2024) Controversies, contradiction, and ‘participation’ in AI. Big Data Soc. https://doi.org/10.1177/20539517241235862 Suchman L (2023) The uncontroversial ‘thingness’ of AI. Big Data Soc. https://doi.org/10.1177/20539517231206794 Valderrama Barragán M, Tironi M, Cotoras D, Correa T, Humeres M, López C (2026) From industry hype to emerging criticism: analysing Chilean news media coverage of artificial intelligence. Digit J 14(2):238–260. https://doi.org/10.1080/21670811.2025.2450624 Funding A.R. acknowledges funding from the ObvIA project (Observatoire des usages, voix et imaginaires de l'IA dans les industries culturelles et la création artistique), supported by the inIdEx ICCA LabEx through the 2025 Structuring Call (AAP Structurant 2025). The inIdEx ICCA LabEx is funded as part of France 2030 and Université Paris Cité's IdEx program (ANR-18-IDEX-0001). Author information Authors and Affiliations Contributions AR, CBE, and JM contributed equally to the conception of the study, the development of the theoretical framework, and the formulation of the conclusions. AR primarily drafted Part 1 of the manuscript, while CBE and JM primarily drafted Part 2. All authors contributed to the writing, revision, and final approval of the manuscript. Corresponding author Ethics declarations Conflict of interest The authors declare no competing interests. Additional information Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Rights and permissions Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. About this article Cite this article Romele, A., Bert-Erboul, C. & Mésangeau, J. Artificial intelligence as political antagonism: media traces of a displaced conflict. AI & Soc (2026). https://doi.org/10.1007/s00146-026-03303-3 Received: Accepted: Published: Version of record: DOI: https://doi.org/10.1007/s00146-026-03303-3