Cultural Imaginaries and Framings of Generative AI
The papers filed here trace a single problem across many surfaces: generative AI does not arrive in the world as a stable technical object but is continually narrated into being — by newsrooms, by industry actors, by media theorists, by creators, and by audiences trying to decide what to trust. Read together, they suggest that “the imaginary” is not a superstructure sitting atop the technology but a set of material practices — sourcing routines, community geographies, discourse genres, even aesthetic conventions baked into models — that co-produce what generative AI is for different publics.
Journalism as an imaginary-making machine
A cluster of papers treats news media as a primary site where GenAI’s meaning is negotiated rather than merely reported. Nguyen2026-vm offers the most granular empirical account, showing that mentalistic-agentic framings of LLMs are real but far from hegemonic in global coverage, and that their distribution tracks editorial style and outlet more than region or culture — a finding that complicates simpler “hype” narratives. Wang2025-zy pushes the comparative-national angle further, operationalizing Cave and Dihal’s hopes/fears framework to map utopian and dystopian tropes across UK, US, Chinese, and Indian newspapers, arguing that these imaginaries are inflected by distinct political-economic stakes in each country. Dodds2026-df reframes the whole enterprise: AI hype, in its account, is not a discursive bubble to be debunked but a systemic infrastructure embedded in journalism’s professional practice, sourcing patterns, and geopolitics — journalists are simultaneously hype’s producers and its watchdogs. Weinbrand2026-sf narrows the lens to a single flashpoint, Google’s AI Overviews, showing how Google, journalists, and SEO marketers each construct competing frames of trust and authority, with the user’s voice and environmental costs conspicuously absent from all of them — a silence that echoes Dodds’s point about hype obscuring hidden costs and labor. Together these papers suggest journalism is not a neutral conduit for AI imaginaries but an active, structurally biased participant in fixing them.
Downstream of these framings sits the question of audience response. Mattis2026-gu, Sbaraini-Fontes2026-cw, and Suk2026-ai form a mini-arc on trust: Suk proposes a multilevel theoretical scaffold borrowing from mass-media and social-media trust research; Mattis empirically shows that all task-specific AI disclosures in journalism depress trust, with fact-checking hit hardest, complicating simple transparency mandates; Sbaraini-Fontes finds, intriguingly, that AI-mediated search is trusted more than AI-produced news, suggesting the “configuration” of AI’s appearance — as infrastructure versus as authored content — matters as much as its presence. Dierickx2026-tw supplies the epistemological substrate for this whole discussion, arguing that existing fact categories (evidentiary, interpretive, institutional) cannot capture GenAI’s probabilistic outputs, and proposing “emergent facts” as a new evaluative category — a conceptual move that helps explain why audiences and fact-checkers alike find AI outputs so hard to trust or discredit cleanly.
Industry discourse and the geography of futuring
A second cluster examines how the AI industry itself narrates its own significance. Stanusch2026-ec dissects the Sam Altman controversy as a moment when industry imaginaries — Longtermism, Regulatory Ambivalence, Techno-Hagiography — surfaced and were defended through “premediation” and “preclusion,” strategies that displace present harms into speculative futures or absorb critique by claiming unique competence to resolve it. Hepp2026-oi complements this with an ethnographic account of Silicon Valley’s “quiet futuring”: beneath the loud CEO pronouncements analyzed by Stanusch lies a fragmented landscape of labs, EA/Rationalist communities, and appropriated places (AGI House, Lighthaven) that supplies the ideational reservoir loud futuring later condenses — both papers converge on the insight that AI imaginaries are not simply top-down PR but emerge from specific social figurations with real geographic and institutional anchors. Richter2026-bt historicizes this terrain, showing that the industry-dominated discourse Stanusch and Hepp describe was not inevitable: US Twitter discourse drifted from institutional variety toward individual tech-influencer dominance only gradually across 2012–2021, while Germany institutionalized in the opposite direction — a divergence the authors argue conditions each country’s subsequent GenAI regulatory posture. Anicker2024-vp provides a theoretical undercurrent for all of this, reframing “agency” itself as a socially granted status rather than an intrinsic property — a move that clarifies why so much industry and journalistic discourse is a contest over whether to grant AI systems agentic standing at all.
Beyond hype: propaganda, epistemic security, and platform transformation
Several papers extend the imaginary lens toward its geopolitical and infrastructural stakes. Beacken2026-zb resists deterministic narratives of AI’s uniform global impact, showing that propaganda uses of GenAI in six democratically weakened states are shaped by local structural conditions rather than technology alone. Emilio2026-ik and Triedman2025-uy both push past the “deepfake” framing toward systemic accounts of epistemic erosion: Ferrara’s “synthetic reality” stack formalizes how fabricated content, identity, and interaction compound into institutional-level trust collapse, while Triedman’s empirical dissection of Grokipedia offers a concrete case of AI-mediated knowledge production diverging ideologically from its crowdsourced source — a Musk-inflected imaginary of “encyclopedic” AI authority materialized in citation patterns. Goldberg2026-eb takes the more optimistic register, imagining LLMs as a potential second paradigm shift for democratic public squares, though its own catalogue of risks (synthetic participation, gameable bridging systems) shows how thin the line is between the deliberative and dystopian imaginaries mapped by Wang. Tornberg2026-lc offers the broadest structural claim: that “social media” itself is dissolving into algorithmic broadcasting, semi-private enclaves, and AI-mediated communication, with generative AI’s synthetic content severing the platform-user link that grounded two decades of media theory — effectively arguing that the imaginary shift documented elsewhere in this set reflects a genuine infrastructural rupture, not just a change in framing.
Generative AI as medium and as culture
A final pair of papers turns from discourse about AI to AI’s own cultural and formal texture. Manovich2026-ih brackets ethical critique entirely to ask what kind of medium generative AI is, arguing it is the first artistic medium whose cognitive capacity — an encyclopedic “media cognition” of art history — is built in, yet one that is structurally conservative and entangles style with content. This medium-theoretic register finds its vernacular counterpart in Galip2026-ix and Tang2026-gu, which examine the aesthetic and labor forms generative AI actually produces at scale: “slop” and “brainrot” as Ngai-esque gimmicks that work simultaneously too little and too hard, and Chinese “AI slop” creators who industrialize virality into an engineered, repeatable craft. Where Manovich finds an idealized, encyclopedic conservatism, Galip and Tang find its degraded, hyper-commercial double — together suggesting that the cultural imaginary of generative AI spans a spectrum from reverent medium-theory to the debased, contradictory textures of everyday platform content, with the “quiet” and “loud” futuring of Hepp and Stanusch operating as the discursive scaffolding that holds both ends of that spectrum together.