Abstract
Generative artificial intelligence has become embedded infrastructure across creative platforms, yet scholarship continues to treat AI primarily as a creative tool or co-creative agent, without interrogating the structural conditions under which creative work unfolds. This article introduces the concept of algorithmic experience governance (AEG): the upstream structuring of creative participation, visibility, sequencing, evaluation and meaning-making through AI-driven platform systems. Drawing on platform ecosystem theory, institutional theory and socio-technical systems theory, the article develops an integrated framework that identifies five governance mechanisms, that is, algorithmic curation, sequencing and path dependency, personalisation as steering, generative mediation and evaluative framing, and maps them across individual, community, platform and ecosystem levels. The framework is grounded in the communicative dynamics of contemporary platforms, including TikTok’s narrative formats, influencer storytelling practices, AI-mediated branding, meme circulation and the cultural semiotics of AI-generated content. Three governance paradoxes and three recursive feedback loops capture the tensions and dynamics inherent in algorithmic governance of creative experience. Four formal propositions reframe creativity under AEG as structurally situated rather than diminished, and a research agenda of 20 questions offers a programme for empirical investigation. The article argues that creativity in AI-driven platforms is not merely collaborative but governed collaboratively and that understanding this distinction is central to reimagining theory and practice.
Introduction
Generative artificial intelligence has moved from experimental novelty to embedded infrastructure. Musicians compose with algorithmic collaborators such as Suno and Amper Music (Kirkpatrick, 2023; Lee, 2022). Designers iterate through generative prototypes on Midjourney and Stable Diffusion (Cetinic & She, 2022; Moruzzi, 2025). Large language models are deployed across journalism, marketing and content production at an unprecedented scale (Ali et al., 2023; Anantrasirichai & Bull, 2021; Holzner et al., 2025). What was once a philosophical question, that is, can machines be creative? (Boden, 2004), has become an everyday reality shaping the creative lives of millions.
The prevailing scholarly response has been to examine AI as a creative tool or co-creative agent. Meta-analytic evidence shows that humans collaborating with generative AI outperform those working alone, though often at the cost of reduced idea diversity (Doshi & Hauser, 2024; Holzner et al., 2025). Others have examined whether AI outputs qualify as genuinely creative or represent what Runco (2023) terms ‘artificial creativity’—original and effective yet lacking intentionality and emergence. Meanwhile, scholarship on creative labour has highlighted how AI dissociates creativity from human agency, raising questions about authorship and the political economy of cultural production (Cloonan & Williamson, 2023; Lee, 2022).
These contributions are valuable, yet they share a common limitation: they treat AI primarily as an instrument that enhances, threatens or transforms creative work, without interrogating the structural conditions under which that work unfolds. Consider a musician releasing an album on Spotify. Her creative process may be entirely human, yet the conditions of her work’s reception are thoroughly algorithmic. Whether her songs appear on Discover Weekly or Release Radar, which of Spotify’s 100 million-plus tracks they are sequenced alongside and how the platform’s engagement metrics assess her ‘success’ are governed by algorithmic systems she neither controls nor fully understands. Her creative agency is real, but it is exercised within an environment that has been structured upstream—before any listener presses play. On TikTok, a similar dynamic operates with even greater intensity: the For You Page algorithm determines within seconds whether a creator’s content enters circulation or vanishes, training creators to adapt their narrative formats, visual aesthetics and emotional registers to algorithmic preferences that are themselves opaque and shifting.
On platforms such as Spotify, TikTok, YouTube and Instagram, AI does not merely assist creativity; it governs the environment within which creative participation, visibility and meaning emerge. The algorithm that determines whether a TikTok video reaches the For You Page, whether an AI-generated brand image gains traction on Instagram or whether a meme format circulates or dies is not a neutral distribution mechanism; it is a governance architecture that shapes what is created, who sees it and what it comes to mean.
This article introduces the concept of AEG to theorise this structural layer. AEG refers to the upstream structuring of creative participation, visibility, sequencing, evaluation and meaning-making through AI-driven platform systems. Drawing on the platform ecosystem theory, institutional theory and socio-technical systems theory, the article develops an integrated framework that identifies five governance mechanisms, maps them across four analytical levels and articulates three paradoxes that arise from their interaction. The article offers three contributions: (a) it defines AEG as a novel construct that integrates previously separate conversations about AI creativity, co-creation and platform governance; (b) it provides a mechanism-level framework grounded in the communicative dynamics of real platforms, that is, TikTok’s narrative formats, influencer storytelling practices, AI-mediated branding, meme circulation and the cultural semiotics of AI-generated content; and (c) it advances a structured research agenda for investigating how algorithmic governance reshapes creative experience across communicative and cultural contexts.
AI and Creativity in Platform Contexts
Scholarship on AI and creativity is organised around three largely separate conversations that together reveal an integrative gap. The first stream positions generative AI as an augmentation tool. Studies span diverse creative domains: design ideation where students using text-to-image AI tools produce outputs rated higher on novelty and elaboration (Chandrasekera et al., 2025), algorithmic music composition that blurs the frontier between social and computational creativity (Ackerman & Loker, 2017; Esling & Devis, 2020) and collaborative business ideation using transformer-based language models (Bouschery et al., 2023). Holzner et al.’s (2025) meta-analysis of 28 studies (m = 8,214 participants) provides the most comprehensive assessment: human–AI collaboration significantly improves creative performance (g = 0.27) but significantly reduces idea diversity (g = −0.86). Ivcevic and Grandinetti (2024) map these dynamics across Kaufman and Beghetto’s four-level creativity continuum, showing that AI’s effects differ from mini-c personal insight through to Pro-C professional creativity. Others question whether AI outputs qualify as creative at all: Runco (2023) argues that they constitute ‘artificial creativity’ or pseudo-creativity lacking intentionality and authenticity, while Moruzzi (2025) reaches a similar conclusion from a process-first perspective that privileges agency and value over product attributes. The limitation of this stream is its framing: by treating AI as an instrument under human direction, it does not interrogate how AI systems shape the creative environment within which that instrument is used.
The second stream examines AI-enabled co-creation. O’Toole and Horvát (2024) review approaches to human–AI collaborative systems spanning ideation support, personalisation, steerable creative tools and the generation of creative constraints. The mixed-initiative co-creativity paradigm explores how human and computational agents contribute to creative content in synergy (Ali & Ali, 2026; Moruzzi, 2025; Yannakakis et al., 2014). Lee (2022) documents how the idea of ‘everyday creativity’ has democratised cultural engagement, decoupling creative practice from professional expertise. Chateau et al. (2025) extend this globally, arguing that Western scholarship dominates conversations about AI and creativity while neglecting the perspectives of the Global Majority. Their 6R framework, including recognition, resituate, remix, resistance, regenerate and reimagine, calls for decolonising approaches to creative AI. This stream, however, tends to take participatory openness for granted. It does not examine how the algorithmic architectures of platforms determine in advance which creators gain access, which content becomes visible, and which forms of creative work are rewarded.
The third stream draws on a substantial body of work examining platform governance and algorithmic power. Gillespie (2014) established that algorithms are not neutral technical instruments but socially consequential systems that structure public discourse through their selection and prioritisation logics, arguing that their apparent objectivity masks deeply embedded editorial choices. Bucher (2018) extended this analysis through the concept of ‘programmed sociality’, showing how algorithms shape not only what users see but also how they feel about and respond to their visibility and invisibility within platform architectures. Pasquale (2015) demonstrated the opacity of algorithmic systems and the power asymmetries this opacity produces, while Ananny and Crawford (2018) argued that transparency alone is insufficient, calling instead for accountability frameworks that examine the broader socio-technical assemblages within which algorithms operate. In the specific domain of media diversity, Helberger (2019) and Napoli (2019) have analysed how recommendation systems reshape the conditions of content exposure, raising concerns about algorithmic monocultures in cultural consumption. Building on this foundation, Rader and Gray (2015) demonstrate empirically that most users have limited awareness of how algorithmic curation shapes their experience and that user behaviour is both input to and output of algorithmic decision-making, a feedback loop with consequences for the system as a whole. Platform governance research documents how recommendation systems shape cultural consumption within the attention economy (Webster, 2014), how AI algorithms reinforce popular cultural tastes at the expense of diversity (Jin, 2021; Kulesz, 2018) and how the hidden labour involved in training and moderating AI systems introduces exploitative dynamics (Chateau et al., 2025). The limitation is that governance is examined separately from creative processes, as a structural phenomenon with economic and political consequences, but not as a constitutive condition of creative practice.
AEG builds on this body of work but departs from it in a specific respect. Existing scholarship on algorithmic governance primarily examines how platforms govern user behaviour, information access and market dynamics (Gillespie, 2014; Gorwa, 2019). AEG narrows the analytical focus to the governance of creative experience specifically, theorising how the five mechanisms identified in this article structure not merely what users encounter but also how creative participation, authorship, evaluation and cultural meaning are constituted within platform architectures. The construct is thus not a relabelling of algorithmic governance but a domain-specific theorisation of its operation within creative ecosystems. Table 1 summarises how AEG relates to and extends three adjacent constructs.
Distinguishing AEG from Related Constructs.
Theoretical Foundations
AEG draws on three complementary theoretical traditions, including platform ecosystem theory, institutional theory and socio-technical systems theory, each contributing a distinct analytical dimension (see Figure 1).
Theoretical Foundations of AEG.
The platform ecosystem theory conceptualises digital platforms as multi-sided intermediaries whose governance function extends beyond market-making to the structuring of visibility itself (Gawer, 2014; Parker et al., 2016). Platforms govern through boundary resources, the technical and institutional interfaces managing complementor relationships (Ghazawneh & Henfridsson, 2013). In creative contexts, recommendation engines and ranking algorithms function as boundary resources that allocate the most consequential resource any creator faces: audience attention (Rai et al., 2019; Webster, 2014). The platform ecosystem theory thus explains governance occurs in the structural architecture of visibility hierarchies, incentive design and algorithmic market-making. However, it cannot explain why specific forms of creativity become valued or marginalised within these structures.
The institutional theory addresses the normative gap. It explains how shared understandings of legitimacy, value and appropriateness emerge within organisational fields (DiMaggio & Powell, 1983; Scott, 2014; Tesluk et al., 1997). When Spotify’s recommendation engine privileges certain musical attributes, that is, tempo, energy and danceability, it implicitly establishes evaluative logics that creators internalise as production norms and audiences accept as consumption expectations (Jin, 2021; Lee, 2022). Over time, engagement metrics displace aesthetic judgement as the institutional standard for creative success, a shift with particular consequences for Global South creators whose cultural logics may not align with Western-trained algorithms (Chateau et al., 2025). The institutional theory explains what governance produces: the normative authority through which algorithms define what creativity means within the fields they govern.
The socio-technical systems theory provides the agency–structure bridge (Emery & Trist, 1960; Trist & Bamforth, 1951). It reveals that algorithmic governance is not imposed on passive creators but co-produced through the recursive interaction between human creative behaviour and algorithmic systems. On TikTok, a creator’s content choices feed the recommendation algorithm (Kang & Lou, 2022), which structures what other users see, which shapes subsequent creative behaviour in a recursive loop (Esling & Devis, 2020; Rader & Gray, 2015). The socio-technical systems theory explains how governance operates: through the co-constitutive relationship between human agency and algorithmic structure, in which neither can be understood in isolation.
Together, these three lenses ground AEG as simultaneously architectural, normative and relational. Crucially, the three traditions do not merely complement each other; they describe a single recursive process. Platform ecosystem structures (boundary resources, visibility hierarchies, incentive architectures) create the conditions within which institutional norms about creative value form. As creators and audiences repeatedly encounter algorithmically curated content, they develop shared expectations about what good creative work looks like, how success is measured and which formats deserve attention. These expectations are not imposed externally; they are co-produced through the socio-technical interaction between human creative behaviour and algorithmic systems, as creators adapt their practice to perceived algorithmic preferences, generating engagement data that feed back into the recommendation architecture, which in turn reinforces or reshapes the institutional norms. On TikTok, for example, the platform’s interest graph (structure) establishes visibility conditions that over time institutionalise particular narrative formats as community standards (meaning), which creators then internalise and reproduce through their ongoing interaction with the algorithm (agency), further entrenching those standards in the system. AEG names the governance that emerges from this recursive interaction across all three dimensions. It cannot be reduced to any single theoretical tradition because it is produced by their convergence.
Algorithmic Experience Governance: Concept, Mechanisms and Framework
Defining AEG
AEG refers to the structuring of creative participation, visibility, sequencing, evaluation and meaning-making through AI-driven systems that shape how experiences and cultural artefacts are curated, distributed, assessed and interpreted across digital platforms. AEG is not censorship, not mere personalisation and not a synonym for recommendation. It is infrastructural orchestration; the upstream governance of the conditions within which creative experience emerges, before conscious user choice is exercised.
Core Mechanisms
AEG operates through five interconnected mechanisms. Each functions across multiple analytical levels, including individual, community, platform and ecosystem, and produces effects extending beyond content distribution into the structuring of creative agency, community norms and cultural value.
Mechanism 1—Algorithmic curation: Ranking, recommendation and content prioritisation systems determine what is visible, to whom and in what order. On TikTok, the For You Page algorithm selects from millions of competing videos to construct each user’s content stream; on Spotify, Discover Weekly and Release Radar curate from over 100 million tracks. At the individual level, curation shapes what creators believe audiences want. At the community level, it produces visibility hierarchies that stratify creators into algorithmically favoured and marginalised tiers. Consider meme circulation on Instagram and X: the algorithmic amplification of certain formats such as reaction memes, duet templates and trending audio clips concentrates cultural attention around a narrow semiotic vocabulary, while non-conforming creative expressions remain structurally invisible (Chateau et al., 2025; Jin, 2021; Webster, 2014).
Mechanism 2—Sequencing and path dependency: AI structures not merely individual encounters but also entire experiential pathways. Early algorithmic interventions, that is, the first videos shown to a new TikTok user and the initial playlist recommendations on Spotify, create path dependencies that shape downstream creative and consumption trajectories. On TikTok, the For You Page within the first minutes of engagement establishes algorithmic assumptions that persistently channel subsequent exposure. For influencers, this means that a creator’s initial narrative format, whether confession-style storytelling, tutorial content or comedic sketch, locks them into an algorithmic identity that becomes progressively harder to escape, as the algorithm optimises for consistency rather than creative evolution.
Mechanism 3—Personalisation as steering: Personalisation systems optimise for engagement by constructing algorithmically tailored environments that feel self-directed but are structurally guided. The tension between relevance and autonomy is particularly visible in influencer storytelling: creators learn through algorithmic feedback which narrative structures, emotional registers and visual aesthetics generate engagement and progressively adapt their creative practice to match. On YouTube, the recommendation engine’s preference for longer watch times shapes not only what audiences see but also what creators produce, a structural incentive towards specific narrative pacing, thumbnail aesthetics and content formats (Kulesz, 2018; Lee, 2022). Personalisation thereby governs both consumption and production simultaneously.
Mechanism 4—Generative Mediation: AI-produced content enters the creative ecosystem directly, creating hybrid authorship dynamics. In AI-mediated branding, companies deploy generative tools to produce marketing visuals, social media copy and brand narratives at scale, content that competes for attention alongside human-created work without clear markers of its algorithmic origin. On platforms integrating generative AI into the creative workflow, such as Adobe Firefly in design, ChatGPT in copywriting or Suno in music, the boundary between tool-assisted and tool-generated becomes structurally ambiguous (Kirkpatrick, 2023; Moruzzi, 2025). Generative mediation becomes governance, rather than merely adding content supply, under three specifiable conditions: first, when platforms algorithmically privilege AI-native formats, for instance, Spotify’s algorithmically optimised instrumental tracks or TikTok’s AI-generated voice filters, giving generative outputs structural visibility advantages over human-created alternatives; second, when AI-generated content enters circulation without legibility markers, making it impossible for audiences to apply distinct evaluative criteria to human versus machine outputs; and third, when the sheer volume of generative content shifts the baseline against which human creativity is assessed, such that the evaluative categories through which creative value is assigned begin to dissolve (Holzner et al., 2025; Runco, 2023). Under these conditions, M4 does not merely increase content supply for M1 through M5 to govern; it actively restructures the competitive and evaluative environment, functioning as governance in its own right.
Mechanism 5—Evaluative framing: Algorithms govern not only distribution but also evaluation. Engagement metrics such as streams, click-through rates, completion ratios and shares function as algorithmically instantiated evaluative criteria that progressively displace aesthetic, critical or peer-based judgement. On TikTok, creative success is measured in views and shares rather than in artistic merit; the platform’s algorithmic logic rewards spreadability over depth, virality over craft (Chateau et al., 2025). At the community level, these metrics become internalised as shared standards of creative success. At the ecosystem level, they reshape the institutional logics governing creative fields: what the algorithm rewards comes to define what creativity means within those fields.
Multi-level Dynamics and Feedback Loops
These mechanisms do not operate in isolation at discrete levels; they interact across a four-level analytical framework, including individual, community, platform and ecosystem, through recursive feedback dynamics that give AEG its systemic character. Figure 2 presents this framework.
The AEG Framework: Mechanisms, Levels and Governance Paradoxes.
At the individual level, AEG shapes perceived creative agency and autonomy. The musician, discussed in the introductory vignette, does not simply create and distribute; she creates within an environment where algorithmic curation has already structured what audiences expect, personalisation has narrowed the audience she can reach and evaluative metrics define whether her work counts as successful. Her creative self-efficacy is conditioned by the governance architecture before she makes a single artistic choice. The influencer who learns that confessional TikTok content outperforms analytical commentary is experiencing evaluative framing as a direct constraint on creative self-concept (Grilli & Pedota, 2023; Ivcevic & Grandinetti, 2024).
At the community level, AEG produces norm formation and creator stratification. Algorithmically amplified aesthetics become community standards (e.g., the ‘TikTok voice’, the three-second hook, the trending audio format), while creators who diverge from these norms are structurally marginalised. Meme culture provides a striking illustration: the algorithmic circulation of certain meme templates on Instagram and X concentrates cultural participation around a narrow semiotic vocabulary, rewarding format adherence over creative originality. Yet communities also generate resistance: counter-algorithmic practices, genre subversion and platform migration represent collective responses to perceived governance overreach. Chateau et al. (2025) document how Global South creators deploy ‘jugaad’—improvisational creativity within constrained systems—as a form of cultural resistance against Western-centric algorithmic logics.
At the platform level, governance design choices carry direct consequences for millions of participants. How TikTok’s interest graph weights engagement versus diversity, how Spotify balances editorial and algorithmic playlisting and how YouTube’s recommendation engine prioritises watch time over creative range; these architectural decisions shape the creative conditions of entire platform populations. These choices are themselves shaped by competitive dynamics across the ecosystem.
At the ecosystem level, AEG redistributes symbolic capital across creative fields, institutionalises algorithmic authority as a legitimate evaluative force and produces long-term cultural shifts in what creativity means and who gets to define it. The cultural semiotics of AI-generated content, that is, the emerging visual language of Midjourney outputs, the distinctive rhythm of ChatGPT prose and the uncanny consistency of AI-composed music, are beginning to constitute a recognisable aesthetic category, one that exists in tension with established markers of human creative authenticity (Esling & Devis, 2020; Lee, 2022; Runco, 2023).
Three named feedback loops connect these levels dynamically. In the governance–norm cascade, platform-level visibility design shapes community norms about what kinds of creativity are valued, which in turn conditions individual creative ambition and self-efficacy. When TikTok’s algorithm consistently promotes three-second hooks and high-energy visual editing, this platform-level choice cascades into community-level aesthetic norms and ultimately individual-level production decisions. In the evaluative reinforcement loop, ecosystem-level institutionalisation of algorithmic authority reinforces platform-level incentive structures, which shape the evaluative metrics communities use to assess creative work. In the resistance–adaptation cycle, individual awareness of algorithmic governance catalyses community-level resistance, counter-algorithmic practices, genre subversion and platform migration, which may prompt platform-level governance adaptations, feeding into ecosystem-level shifts in institutional logics.
Governance Paradoxes
Three paradoxes emerge from the interaction of AEG’s mechanisms across levels, not as abstract tensions but as structural consequences of the dynamics described above.
The first is the participation paradox: generative mediation (M4) dramatically expands who can participate in creative production, as anyone with access to Suno can compose music and anyone with Midjourney can generate images. Yet algorithmic curation (M1) simultaneously concentrates visibility power, ensuring that expanded participation does not translate into expanded recognition. More people can create than ever before, but fewer algorithmic systems determine whose creativity is culturally consequential. The democratisation of production coexists with the centralisation of governance.
The second is the seamlessness paradox: Personalisation (M3) and generative mediation (M4) reduce friction in creative processes, making production and distribution increasingly seamless. Yet this very seamlessness diminishes the deliberation, critical reflection and effortful engagement that creativity scholarship identifies as central to meaningful creative work (Esling & Devis, 2020; Runco, 2023). The ease with which AI-generated brand content, templated TikTok narratives and algorithmically optimised visuals flood social media platforms intensifies curation’s gatekeeping role: the more content is produced, the more algorithmic governance determines what matters.
The third is the personalisation paradox: Personalisation (M3) increases the perceived relevance of creative content, making the experience feel responsive and self-directed, while evaluative framing (M5) simultaneously obscures the governance architecture that produces this experience. The more tailored the encounter, the less visible the governance, which is what Rader and Gray (2015) identify as a fundamental opacity in which algorithmic structuring operates below the threshold of conscious engagement.
Together, these paradoxes constitute the distinctive normative terrain of AEG: they are not problems to be solved but structural tensions inherent in the algorithmic governance of creative experience.
Reimagining Creativity and Co-creation Under Algorithmic Governance
The AEG framework reframes how creativity and co-creation should be understood in platform environments. Creativity research has long acknowledged that creative action is situated within enabling and constraining contexts (Amabile, 1983; Woodman et al., 1993). Grilli and Pedota (2023) argue that AI may change not only creative performance at individual, group and organisational levels but also the relative importance of enabling conditions themselves. AEG extends this insight: in platform environments, the context is not a passive background but an actively governed architecture. The algorithmic systems that structure visibility, sequence experience, personalise encounter, mediate production and frame evaluation are not external to the creative process; they constitute its operating conditions. This generates four propositions (Table 2).
Summary of Propositions.
The creativity literature has traditionally centred individual cognition: divergent and convergent thinking, domain expertise and intrinsic motivation (Amabile, 1983; Guilford, 1984; Runco & Jaeger, 2012). These remain relevant, but AEG reveals that they operate within a governance architecture that pre-structures their expression. A TikTok creator’s divergent thinking is exercised not in an open field but within an algorithmically shaped environment that has already determined which narrative formats gain traction, which audiences are reachable and what evaluative metrics define success (Esling & Devis, 2020; Holzner et al., 2025). As such, we propose the following:
Proposition 1: Creativity under AEG is not diminished but recontextualised; it shifts from an individual cognitive act to a structurally situated practice embedded within algorithmic infrastructures.
On TikTok, co-creation through duets, stitches and shared audio clips is mediated by recommendation logics that determine which collaborative content enters wide circulation and which remains confined to niche audiences. On Spotify, collaboration between musician and algorithm is shaped by playlist logics that favour certain production qualities. Co-creation under AEG is not free form collaboration but governed collaboration, whose possibilities are pre-structured by the platform’s governance design (O’Toole & Horvát, 2024; Vear & Poltronieri, 2022; Vinchon et al., 2023). As such, we propose the following:
Proposition 2: Co-creation pathways in AI-driven platforms are not open-ended but algorithmically channelled visibility systems, sequencing logics and evaluative metrics pre-structure the collaborative space.
Algorithmic curation and personalisation allocate attention, a finite and consequential resource in the attention economy (Webster, 2014). The result is systematic asymmetry between creators who align with algorithmic logics and those who do not. Chateau et al. (2025) demonstrate that these asymmetries operate along cultural and geopolitical lines, with Global South creators systematically disadvantaged by AI systems trained on Western-centric data and optimised for Western consumption patterns. The sequencing and filtering of creative communication is thus not merely a technical process but a power-laden one with distributive consequences. Therefore, we propose the following:
Proposition 3: Communication flows within creative platforms are not neutral but sequenced and filtered, producing asymmetries in whose creative voice reaches audiences and whose remains invisible.
When engagement metrics function as proxy evaluative criteria, they do not simply measure pre-existing values; they constitute it. A track with high completion rates is not merely popular; it is algorithmically legible as successful, which feeds its further promotion, reinforcing the evaluative standard. Over time, this recursive dynamic reshapes the institutional logics of entire creative fields, what Lee (2022) identifies as the subsumption of aesthetic judgement by market metrics and what Grilli and Pedota (2023) describe as AI’s potential to alter the conditions under which creative work is recognised and valued. Based on this discussion, we propose the following:
Proposition 4: Cultural meaning in AI-mediated environments emerges within governance constraints, not outside them; what counts as creative, original or valuable is partly an algorithmic determination.
Research Agenda
The AEG framework opens a structured space for empirical inquiry. Studying AEG requires methodological pluralism: combining computational methods such as algorithmic audits and trace data analysis with interpretive methods such as qualitative interviews and ethnography, in multi-level longitudinal designs capable of capturing cross-level dynamics. The questions below are derived directly from the mechanisms and paradoxes developed in the fourth section 4, and Table 3 presents 12 research questions organised by the AEG mechanism.
Research Agenda for Algorithmic Experience Governance.
M1: Algorithmic curation; M2: Sequencing and path dependency; M3: Personalisation as steering; M4: Generative mediation; M5: Evaluative framing.
Feedback dynamics reference the governance–norm cascade, evaluative reinforcement loop and resistance–adaptation cycle as defined in the fourth section.
Methodological implementation guidance: ‘Algorithmic audits’ refer to controlled experiments probing platform recommendation systems through test accounts or varied content submissions to observe differential algorithmic outputs across conditions. ‘Trace data analysis’ involves computational analysis of platform data, including engagement metrics, content metadata and temporal patterns of visibility. ‘Computational content analysis’ uses automated feature extraction (visual, textual, auditory) to measure shifts in creative output characteristics across creators or time periods. ‘Format entropy’ quantifies the diversity of content formats within a creator community using information-theoretic measures. ‘Longitudinal discourse analysis’ tracks shifts in evaluative vocabularies used by creators and audiences, examining whether success is increasingly described in algorithmic terms (streams, completion rates) rather than aesthetic ones
For algorithmic curation (M1), the central questions concern how creators perceive and adapt to governance, and whether curation amplifies dominant aesthetics at the expense of diversity, particularly as Holzner et al.’s (2025) individual-level findings aggregate through the governance–norm cascade. For sequencing and path dependency (M2), research should examine how early algorithmic interventions lock creators into algorithmic identities and under what conditions these dependencies can be disrupted. For personalisation as steering (M3), the personalisation paradox generates a testable boundary condition: there may exist a threshold beyond which personalisation shifts from enabling relevance to constraining discovery, moderated by individual dispositions such as creative self-efficacy (Amabile, 1983; Ivcevic & Grandinetti, 2024). For generative mediation (M4), the key question is when AI-generated content constitutes governance rather than merely adding supply and how communities resist or adapt to its effects (Chateau et al., 2025). For evaluative framing (M5), longitudinal studies should track whether engagement metrics progressively displace aesthetic evaluative standards within creative fields. Two cross-mechanism questions address comparative platform analysis and the variation of governance effects across cultural contexts.
To illustrate the tractability of this agenda, consider a concrete study design testing the governance–norm cascade on TikTok. The research would exploit a known algorithmic policy change, such as TikTok’s periodic adjustments to For You Page ranking criteria, as a natural experiment. Computational content analysis of creator output in the months before and after the change would track shifts in video duration, narrative format distribution, audio usage patterns and visual editing styles, while semi-structured interviews with creators across follower tiers would capture how they perceive, interpret and adapt to the changed algorithmic environment. Stratification metrics such as Gini coefficients of view distribution could operationalise the visibility hierarchy dimension, while format entropy measures could capture community-level aesthetic convergence or diversification. If the cascade holds, algorithmic changes at the platform level should produce measurable shifts in community-level content distributions, followed by individual-level strategic adaptations, with timing patterns consistent with a top-down diffusion process. A design of this kind yields falsifiable predictions directly derived from the framework, demonstrating that the AEG agenda is not merely conceptual but empirically actionable.
Conclusion
This article has introduced AEG as a theoretical construct for understanding how AI-driven platforms structure the conditions of creative participation, visibility, evaluation and meaning-making. Drawing on the platform ecosystem theory, institutional theory and socio-technical systems theory, it has argued that algorithmic systems do not merely augment or facilitate creativity; they govern the environment within which creative experience unfolds.
The AEG framework identifies five mechanisms including algorithmic curation, sequencing and path dependency, personalisation as steering, generative mediation and evaluative framing, through which governance operates across individual, community, platform and ecosystem levels. Three paradoxes capture the structural tensions inherent in this governance; three recursive feedback loops reveal its dynamic, co-constitutive character. The framework has been grounded in the communicative realities of contemporary platforms such as TikTok’s narrative formats, influencer storytelling practices, AI-mediated branding, meme circulation and the emerging cultural semiotics of AI-generated content, demonstrating that algorithmic governance is not an abstract structural condition but an everyday shaping force in creative life.
As generative AI becomes more deeply embedded in creative platforms, from Spotify and TikTok to Midjourney and ChatGPT, the governance of creative experience will increasingly determine not only what is produced and distributed but also what is valued, by whom and on whose terms. The 4 propositions and 20 research questions advanced in this article offer a structured programme for investigating these dynamics empirically and across cultural contexts. The question is no longer whether algorithms shape creativity. It is how we understand, investigate and ultimately govern the governance, ensuring that the algorithmic architectures reshaping creative life remain subject to the same critical scrutiny we apply to any consequential form of power.
Footnotes
Authors’ Contribution
Faizan Ali: Supervision, data curation, writing—original draft.
Fahad Mohammed Alhuqbani: Resources, writing—review & editing.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The authors received no financial support for the research, authorship and/or publication of this article.
