Methodological Observations and Challenges

The papers in our database use a range of qualitative methods including qualitative content or discourse analysis (1, 6, 13, 19, 20, 29, 67), interviews (15, 20, 59), case studies (16, 30, 32, 33, 46, 51), focus groups (56), various ethnographic approaches (11, 14, 17), practice-based artistic research (28, 41, 61) and policy or critical analysis (36, 37, 47, 49). There were few papers which also deployed large-scale quantitative methods (such as 68). (17) for example underlines the need for "urgent critical analyses", and the necessity "to formalize the novel kinds of research methods it both demands and facilitates" (De Seta et al. 2024, 2). On the one hand, the absence of training datasets, or the impossibility to access those, is highlighted; on the other hand, the available ones are impossible to study qualitatively due to their scale (17).

In addition to the high number of qualitatively oriented studies, a significant portion of our dataset consists of conceptual papers. These articles engage with the theoretical foundations of AI research by case-based reflection (30, 50) or proposing frameworks for investigation, like for instance posthumanism (8, 42), discourse analysis (13), critical theory (49), combinatorics (55), or Actor-Network-Theory (64). The importance of interdisciplinarity is also emphasised, with a call for researchers to collaborate across various fields to address the cultural, social, and technical complexities that AI presents (23, 34, 47). Essays and commentaries, on the other hand, are often characterised by fundamental philosophical (2), historical (8, 15), or legal considerations (7, 22, 33, 47, 53).

Table 2 shows the distribution of the methodological-conceptual orientation of the articles across the various genres represented in our dataset.

Understandings of Creativity and Aesthetics

A recurring theme is the contested notion of originality in AI-generated art. Several papers highlight that AI outputs are massive in scale (for example, AI composers are reported to be capable of generating 10,000 songs per day, 46) which also leads to de-evaluation of AI generated against human-created art work (3, 14, 19). AI-generated art can also provoke feelings of unease (13). Seen in this light, AI may assist in exploring and interpreting creative options (40, 55), thereby supporting but not truly replacing the creative process. However, many papers discuss that the very concept of authorship is contingent and subject to historical and social transformation (18, 52).

While some argue that the AI output is clearly bound to the training data and cannot be in this sense novel (5, 10) or subversive (19), others refer to society's obsession with originality and state that human-produced art never appears "ex nihilo or from nothing" (Berkowitz 2024, 407). Yet, the unique lived experiences are often regarded as a distinguishing factor to define the value of human created art (3, 59). Artistic identity and agency are central to these discussions (1, 18, 46) and the output of AI models is often evaluated in terms of its aesthetic value (46), novelty and originality (24, 46). The originality of mere AI output is questionable, as it tends toward stylistic uniformity based on algorithmically preferred conventions (12). AI-generated art is perceived as less creative, at least when it is transparent that the content was generated by AI (3). Conversely, AI-generated art is evaluated similarly when the context of its creation is obscured.

There are papers that see AI less as a creator, but more as a tool (33, 66), complementor (63), collaborator (1), or co-creator (18, 24, 51). AI can challenge artists' integrity and reputation (33), particularly when massive volumes of content are generated in the style of specific artists, potentially devaluing the original (53) or when it offers new machine-generated aesthetics (27). AI has contributed to the dehumanisation of creativity by framing it as a form of capital detached from human subjectivity (36). In this context, creativity emerges as neither fully human nor machinic but as a "networked and distributed form of agency" (Chow and Celis Bueno 2025, 2), hedged and formed by the platforms that are being used (51). Some papers showcase how AI's contribution can be framed, considering also human intentions and showcasing a view of creativity shared between human and non-human agents (18, 40, 49, 57), with others extending this perspective toward posthuman aesthetics (46, 58, 60). In many papers the focus is less on whether AI replaces human creativity, and more on questions of human-AI co-creation (18, 45, 51, 52, 56, 57) as well as the integration of machine learning and human creativity (23, 62). So, new creative horizons are possible when human creativity is combined with AI creativity. AI is then more of a performative or creative partner than being granted sole authorship. However, it is more than just a tool; it can be conceived as an improvisational actor (51, 58). The degree of artistic control that is enabled (or not enabled) by interfaces and systems is a subject of consideration, as is the resulting change in the creative role of humans (10). Here it is also interesting to note how many of the papers are examining different implications of AI in the music sector (21, 26, 31, 35, 42, 54, 57, 61, 62, 64, 65, 69¹).

Some papers provided historical overviews (4, 13, 37), tracing the development of AI and its influence on creative industries. We also identified papers that explored paradigm shifts in the cultural and creative sectors (20, 29, 53). One (58) even discusses the aesthetic and ethical implications of sonic collaboration with animals and AI, suggesting that this could broaden aesthetic resources, and extend creative agency beyond the human. AI is giving rise to new aesthetic paradigms, such as the "disconnection" (connectionism) of neural networks as opposed to the linearity of earlier systems (2). The disconnect between what is shown, and its real-world counterpart is undermined by AI-generated images (Midjourney, DALL-E), which poses a challenge to traditional image reception and criticism (13). This undermines the ontology of the image and blurs the distinction between "real" and "virtual" (49).

Numerous potential benefits of AI for creative industries are emphasised, such as revolutionizing customer service, for example through democratisation of the music production (29), cutting costs (36), increasing efficiency, for instance, in screenwriting (59), enhancing creativity (63), providing new perspectives on the audience perception or by suggesting a new artistic approach (14), and reaching new, e.g. younger audiences (15). However, concerns are raised about the unchecked commercialisation of AI-generated music and its implications for the industry, which is seen as problematic in the long term (29). The commercial realities of the industry continue to dictate its direction (59), often at the expense of the creative labour force (5). AI automation tools have been credited with saving considerable time for practitioners, however, that time often has to be dedicated to less creative tasks, such as prompting, checking, and correcting the results of the automation (14). Technical competencies and their application is also an area that the papers covered, discussing how AI works and how it can be applied (see as for example 9, 43, 44, 45). Furthermore, the prevalence of generative AI and how it is here to stay is argued by authors of paper 17 (De Seta et al. 2024, 12), and various concerns about irreversible changes to the industry that AI would bring are vocalised in several papers (e.g., 7, 17, 34, 39, 59), especially in terms of increased competition for certain categories of creative workers (36). Despite these concerns, (48) notes that AI has not yet reached a level of genuinely "smart machine" that could replace humans in creative roles.

Copyright, Legal Considerations and Policy Implications

Copyright is one of the critical points that is discussed in several papers, as current laws were not designed to address AI's role in creativity, and the legal frameworks often fail to protect artists adequately. Authorship and ownership are also discussed in the papers from a legal perspective (8). The ownership of AI-generated works is questioned, and it is not clear if or how these works can be copyrighted at all (33, 47). EU AI Act (59), US Copyright Act and Digital Millennium Copyright Act (DMCA) (33) are also criticised for its westernised nature that dictates the individualised ownership. As laws are concentrated on the protection of the individual, they can leave communities vulnerable to exploitation through data-driven practices (33) or data colonialism (34). The application of the copyright laws also cannot be universal, due to the different legal frameworks, judicial systems, and various factors involved (such as the nature of resulting work, "transformative" use of content, or parody, type of data in the training set, which is also hard to reverse-engineer, effect on potential market, 33). AI models work at the intersection of legal frameworks and economic value (69²) and often sidestep copyright by scraping training datasets without any consent (36). AI also allows for potential exploitation and plagiarism (63).

All this has led to calls for a reevaluation of copyright laws (22, 59), as well as discussions on the legal protection of data used to train these systems (33). The challenge posed by AI to the current copyright system is especially evident in music, where AI-generated covers and imitations of artists may raise legal concerns, for example, regarding deceased artists' voices (24). The appropriation of user data for training AI models, referred to as "data grab" or "datafication" (34) is closely linked to the concept of "surveillance capitalism", where data is extracted and monetised by large tech companies, often without compensating the individuals whose data is used (Zuboff 2020). These power asymmetries deepen inequalities, as large corporations disproportionately benefit from "free data", reinforcing existing societal divides (34).

A central aspect of the debate on AI in cultural production is legal regulations and policies for AI-generated content and where they fail. Authorship and originality must also be renegotiated in relation to legal issues, as IP rights have so far been based on human authorship, which can no longer be upheld in the case of AI (22). It is therefore clear that new technical procedures such as generative licensing must be found in order to renegotiate issues of contract design and remuneration (47). One problem here is that AI is handled differently internationally, and it is difficult to enforce uniform global standards due to differing interests and due to dynamic technological developments.

The abundance of algorithmically curated content and personalised delivery mechanisms shift the balance of power between creators, platforms, and audiences (59). This change is possibly showing a broader societal shift from a creator-centered approach to an audience-centered model (46), where the focus is on curated consumption rather than production. The platformisation (Nieborg, Poell and van Dijck 2023) of creative work in this context has further raised concerns about the erosion of autonomy for individual creators, with algorithms governing much of the visibility and distribution of their work. This challenge is further complicated by the persistence of algorithmic black box systems that obscure decision-making processes (16, 17, 27) and raising concerns about accountability and transparency.

Ethics

The ethical discussions on AI in the creative industries are not unique to this sector but reflect broader debates on the ethics of AI in society. Generally, the authors raise fundamental questions about authorship (24), accountability (18, 27), responsibility (5), ownership (59) and representation (25, 40) that have already been distilled in the literature on AI ethics.

Concerns about privacy and security are present, particularly regarding the risk of data breaches, misuse, and hacking (5), as well as the exploitation of large volumes of user data for personalisation of content (29). AI is also connected to the issues of political misinformation, deepfakes, and other harmful contents, as AI allows for the manipulation of public opinion (17). AI's societal impact is also tied to questions of bias, as machine learning models can reflect and reproduce racial, gender, and social stereotypes (27), deepening the systemic inequalities and polarisation (48). Representation of entire cultural groups is also an issue, as AI can marginalize or make invisible cultural groups just by excluding them from training data (25).

While the papers are quite diverse in theoretical and conceptual approach, basically all of them touch on ethical questions concerning the use of artificial intelligence in artistic and cultural contexts. One important area of reflection is the shifting notion of authorship and agency. Several contributions (e.g., papers 2, 10, 18, 47, 49, 55, 58) explore how AI challenges the idea of the individual, intentional human creator. The capability of AI to serve as a tool for collective creativity further challenges the boundaries of authorship, making it a shared process among machines, humans, and the public (24). Some papers adopt a posthumanist and post-anthropocene (1, 42, 46, 58, 60) perspective, considering how creative agency can be extended beyond humans also to AI actors. Some are also tackling the issue of the extent of human accountability and responsibility in the creative process (56) and the development of suitable ethical frameworks.

A second cluster of ethical concern revolves around aesthetic and moral judgement in relation to AI-generated artefacts. In particular, paper 3 demonstrates empirically how knowledge about a work's AI authorship influences moral evaluations and perceived artistic value, in most cases evaluating the AI content lower than human-made art works.

Interfaces and design are also the subject of ethical considerations (10), as the design of user interfaces influences the extent to which creative agency is either supported or restricted, and the extent to which decisions are taken or allowed. Accessibility and inclusion are also influenced by design and interfaces (38, 52). AI is not neutral, but a tool that can favour certain practices or communities over others. Therefore, keeping control and ensuring human agency remains a central concern in the development and deployment of AI (59). There is a certain threat that important human decisions could be substituted by algorithms, which is seen as not only "ethically dubious but also technically premature" (Rindzevičiūtė 2022, 832). The notion of Human-in-the-Loop (59) or even Musician-in-the-Loop (34) describe the necessity of human oversight, creative judgement and centrality of human agency for the operation of AI systems.

There is also a call for responsibility among researchers and developers to engage with communities of practice, ensuring that the values of these communities are respected while introducing transformative technologies like AI (34). It is also crucial that research on AI not solely come from the Global North, as pointed out by (11, 17, 40), to ensure more diverse perspectives in AI development and application.

Overall, the authors call for greater responsibility in the development of AI technologies and emphasise the need for transparency, traceability, ethical dataset curation (53), and fairness (27). Companies are urged to consider the cultural implications of their AI tools, ensuring that these technologies do not misrepresent or harm the values of the communities they affect (34). There is also a growing demand for policies to mitigate the inequality of access to AI tools, particularly in the context of global disparities in technology access (48). The policy conversation must evolve to address these issues in ways that protect individuals, respect cultural contexts, and foster social inclusion while challenging existing power structures (34, 53).

Methodological Reflexivity

Our methodological observations imply that researchers examine AI functions from different perspectives, considering not only what it does but also how it works and how it can be critically assessed. Our study reveals the topic of AI has led to the production of a lot of exploratory, qualitative and conceptual papers. AI is not examined in research just as a technical tool, but as a development that raises complex questions on creativity, authorship, and ethics, that are not easily addressed by a singular discipline. Papers are coming from a variety of different disciplines, showcasing the penetration of AI in different academic areas, and the consideration given by researchers on the role it already does and will play in the future.

Understanding early phenomena makes the use of perspectives from multiple fields, such as for example here cultural studies, heritage studies, media studies, or computer science inevitable. The diversity of themes highlights how this interdisciplinarity is necessary in the studying of AI to fully grasp its effects and implications. The reviewed papers reflect this, touching on both technical/methodological and philosophical/ethical considerations. Scholarship is thus not limited to questions of functionality, but also addresses the ways in which it transforms human values, practices, and institutions, signaling a consideration of balance between technical and humanistic concerns.

Authors are also engaging with a variety of theoretical frameworks and use different theoretical approaches. We see this as a finding per se, as it showcases how research in the area is still in an exploratory stage, signifying the start of a new domain for scholarship. There is not much empirical research conducted yet, and there are particularly few quantitative studies, but the topic of AI seems to gradually occupy more space in the scholarship of the CCS. The different methods used in the papers may be rich in insight but may sometimes lack scalability and generalisability, which is not surprising given the research on the various implications of the introduction to AI is still new.

In our case, the fragmentation of the studied literature showcases on one hand the interest in the topic by many disciplines that do not necessarily belong to the same area (e.g. social sciences and humanities), and on the other hand the lack of a common conceptual framework or shared terminology that would allow for more dialogue across these fields. This disciplinary fragmentation results in parallel discourses that can overlook each other's insights, thereby limiting the potential for a more comprehensive understanding of the impact of AI in cultural production. Publication of more research in the topic will probably eventually bridge these gaps and allow for even more synthetic research and interpretations.

Centrality of Creativity and Aesthetics

Across the papers, the profound re-imagination of creativity is a recurring theme. Different understandings of creativity and aesthetics in the pool of papers underline how research on AI questions traditional notions of creation, originality and artistic agency, showcasing a cultural shift in the definition and evaluation of creativity in cases where non-human agency is involved. New meanings of creativity including human and AI collaboration and how they affect and redefine the creative process are discussed, as well as new considerations on aesthetics that include non-human agency or even substitute human creativity. AI does not only serve a tool, but is increasingly positioned as a collaborator, a co-creator, and sometimes even as a disruptor of traditional artistic paradigms. However, this shift brings forward critical questions of authorship, legal frameworks, and the question of how existing research methodologies can capture the still fluid, interdisciplinary processes that AI comes hand-in-hand with.

Music is one of the sectors that is seen to be very much related with AI (see Table 2), for a set of different possible reasons. One is the association of the music sector with technology, which inevitably makes AI a significant factor that is expected to trigger change and developments in the way music is created. The other, associated with the first, is the implications of these technological advancements in music production. This is directly related to the fast speed and low costs of the integration of AI in key parts of the value chain in the music sector, and the financial advantages this is expected to have.

Legal and Policy Urgency

The introduction of AI in culture and the integration of non-human agency in the production of culture has implications on copyright, legal issues and working conditions, and brings to surface policy considerations in the form of a growing urgency to address this integration regarding authorship and the rights of artists and users. In the cultural sector, and particularly in the music sector, issues regarding working conditions are constantly in the research agenda of scholars (Kolokytha et al. 2025). The introduction and application of AI makes the sector more vulnerable to uncertainty as creative and intellectual labour is highly affected by AI.

On the one hand, AI is seen as a democratising factor which allows for more public participation, active co-creation, and provides professional-level tools to those that could not otherwise afford them. In this sense, it continues the trajectory of digital tools in general (Shirky 2008, 55-80), but with a stronger impact: it does not merely facilitate creativity (e.g. Holst et al. 2025), it can, at least in parts, replace it. AI appears to be very promising, especially in terms of efficiency and cost-reduction particularly of human labour, although so far real-world applications are limited. On the other hand, its use can lead to increasing existing inequalities and can affect access to culture by those who do not have the education, training or competences to keep up with technological developments (Dada et al. 2025).

Particularly in the creative sector where the human experience, individuality, emotionality and the human agency are valued by the audiences and are shaping not only the experience but the economic worth of cultural goods, the use of AI highlights a scholarly problematic that revolves additionally around issues of authenticity, human creativity and artistic integrity. Despite its immense ability and potential to generate content, AI lacks the human/lived experience that is part of the merit, value and meaning of cultural goods and experiences which ultimately makes them resonate with audiences and consumers. This, in turn, challenges long-established notions of authorship and poses questions of legitimacy of AI-generated works.

Ethical Complexity

Ethical concerns revolve around bias, accountability and power, and demonstrate how AI in culture is not about its mere use as a technical means, but as an element that has repercussions on values, inequalities and societal norms. AI sparks a lot of ethical discussions, especially when it comes to issues regarding authorship, copyright law, important decision-making and transparency. AI development often relies on vast amounts of data sourced from around the world, leading to issues of data colonialism (Kwet 2019; Pinto 2018) and raising the questions of fair representation of non-western perspectives in both the datasets used, and the narratives generated by AI systems.

There are papers that address a broader posthumanist and post-anthropocene debate reconsidering the role of the automation and non-human collaborators, and distributed agency. The increased presence of AI puts human artists as well as marginalised voices in danger and may enable a concentration of creative and even economic power to those who control technology or are digitally literate and raise precarity for those who are not.

Some pressing ethical issues are not being discussed in the literature, although they occur in the public debate. In particular, concerns about the copyrighted and creative materials for training AI systems are hardly raised in the papers. The same is true with regard to labour markets, legal protection of artists intellectual property rights and their remuneration. In this respect, the literature reviewed is mostly focused on symbolic and design-oriented concerns, while broader systemic issues that look at the cultural economic aspects such as data ethics and platform regulation are not brought up.

Across our identified areas of research, scholars reflect both optimism and concern. On one hand, AI is seen as an enabler of new forms of creativity, access, and participation; on the other, scholars warn of biases in training data, opaque algorithmic decision-making, and the risk of diminishing human agency. Importantly, many studies highlight the need for more inclusive, critical, and context-sensitive approaches to AI adoption in creative environments.