Online commercial artificial intelligence (AI) platforms for generating music from text prompts (AI music) are now being used by many users to create millions of music audio recordings daily. Some AI music is appearing in advertising, music playlists of restaurants and gyms, and even hit music charts, in many countries. How are users engaging with these text‑to‑music AI platforms, where text is a principal mode of interaction to specify prompts (e.g., free terms), lyrics (e.g., sung terms), and tags (e.g., high‑level stylistic terms)? What languages appear? What characterizes prompts, lyrics, and tags? How are mentions of real artists used? What kind of additional instructions (metatags) are used? To address these questions, we assemble and analyze a collection of 101, 953 songs generated from May to October 2024 by 60, 342 users of Suno and Udio. Using a combination of state‑of‑the‑art text‑embedding models, dimensionality reduction, and clustering methods, we analyze the prompts, tags, and lyrics and automatically annotate and display the processed data in interactive plots. Our results reveal prominent themes in lyrics, language preferences, and prompting strategies, as well as peculiar attempts at steering models through the use of metatags. We share our code and data resources to promote further musicological study of AI music.

The impact of AI technologies on cultural and creative sectors and industries is expected to increase dramatically in the upcoming years. As significant uncertainties rattle the political economy of artmaking, not all stakeholders feel their interests and concerns are equally recognised. At stake here are not only individual career prospects but the entire horizon of our envisioned cultural-technological futures. 

This dissertation examines artificial intelligence (AI) models in creative domains, focusing on their emerging impact on the conditions and infrastructures of artistic work. It frames the transition of this socio-technical system as an instance of a broken world, and extends Creative AI ethics beyond abstract principles towards the situated, lived experiences of the various stakeholders involved: artists and music communities, as well as developers and service providers of Creative AI applications. Grounded on critical analysis and ethics of care, the dissertation explores the tensions and transitions practising artists experience in their working environments in the context of rapid AI proliferation, and how AI development work practices could better support artists in navigating the rapidly changing socio-technical AI landscape. 

Drawing on six empirical case studies, the dissertation contributes, first, an analysis of the frictions and reconfigured conditions of artmaking, as experienced by Nordic AI artists in the early 2020s. Second, it introduces care ethics to the analysis of Creative AI and proposes sustained reflexive repair as a novel conceptual lens for managing disrupted cultural data relations in AI development. Overall, the dissertation argues that foregrounding repair can make the emergent ethical impacts of AI technologies and practices visible and help reconfigure paradigms for developing, designing, using, and regulating Creative AI that foster fair market conditions for artist economies and cultivate artistic integrity and agency.

Påverkan av artificiell intelligens (AI) på de kulturella och kreativa sektorer och industrier förväntas öka dramatiskt under de kommande åren. Medan betydande osäkerheter skakar den politiska ekonomin kring konstnärligt skapande, upplever inte alla aktörer att deras intressen och bekymmer tas på lika stort allvar. På spel står inte bara individuella karriärmöjligheter, utan även hur vi föreställer oss vår kulturteknologiska framtid.

Denna avhandling undersöker AI inom kreativa områden, med fokus på dess framväxande påverkan på villkoren och infrastrukturen för konstnärligt arbete. Genom att tolka denna sociotekniska omvandling som ett exempel på en “trasig värld” (broken world), utvidgar avhandlingen den kreativ AI etik bortom abstrakta principer till de situerade och levda erfarenheterna hos de olika inblandade aktörerna: konstnärer och musikgemenskaper, samt utvecklare och tjänsteleverantörer av kreativa AI-applikationer. Avhandlingen grundar sig på kritisk analys och omsorgsetik. Den utforskar spänningar och övergångar som yrkesverksamma konstnärer upplever i sina arbetsmiljöer i samband med den snabba spridningen av AI samt hur arbetspraktiker inom AI-utveckling bättre kan stödja konstnärer i att navigera i det högst dynamiska sociotekniska AI-landskapet.

Här bidrar sex empiriska fallstudier för det första med en analys av de friktioner och omformade villkor för konstnärligt skapande så som nordiska AI-konstnärer upplevde under det tidiga 2020-talet. För det andra introducerar avhandlingen omsorgsetiken i analysen av kreativ AI och föreslår hållbar reflexiv reparation (sustained reflexive repair) som en ny begreppsram för de splittrade kulturella datarelationerna i AI-utveckling. På en allmän nivå argumenterar avhandlingen att ett fokus på reparation kan synliggöra de framväxande etiska effekterna av AI-teknologier och -praktiker, samt bidra till omställda paradigm för hur kreativ AI utvecklas, designas, används och regleras på ett sätt som främjar rättvisa marknadsvillkor för konstnärligt ekonomi och stärker konstnärlig integritet och medverkan.

The fast proliferation of AI-driven singing voice synthesis (SVS) facilitates wide-scale misappropriation of third-party vocal expressions, while sidelining the interests of practicing vocal artists. Since EU law provides limited protection for the human singing voice, the emerging SVS industry practices and norms are mainly self-regulated through the Terms of Service (ToS). In this study, we survey the ToS of ten commercial SVS services to examine how they respond to the intensifying calls for Responsible AI and articulate their understanding of a fair market. This is the first paper that examines the contractual dispositions of the service providers of generative AI models and their ethical implications in generative music applications for singing. Using the lenses of procedural and substantive fairness, we chart how service providers allocate rights and obligations among the stakeholders and outline the forms of agency the voice artist can be assigned despite the legislative void. Our work contributes to the socio-technical study of fairness in the context of the developing SVS market. It also informs generative AI policies and further advocacy efforts to cultivate responsible innovation and defend the integrity of vocal artists against synthetic exploitations.

Adoption of AI technologies in the arts has faced a divided reception and raised controversies—while some artists critique them, others embrace the opportunities they provide. In this paper, we contribute empirical insights into what motivates artists’ adoption of AI art technologies, as well as what kind of worldviews these motivations for AI adoption in artistic contexts embody. We shed light on these questions by presenting results from a qualitative interview study (N = 19) of AI artists practicing in Nordic countries. Five key themes emerged that describe artists’ motivations for using AI in their work, which we analyse in reflection with Kerschner and Ehlers' Attitudes Towards Technology (ATT) framework in terms of underlying attitudes of AI-Enthusiasm, Pragmatism, Romanticism, and Scepticism. Our analysis diversifies the understanding of why artists use AI, surfacing cultural and political tensions inherent in adoption of AI technologies for artistic and creative use.

The advancement of AI-tools for musical performance has inspired exciting opportunities for interaction with musical-AI-agents. Interactions between humans and AI-agents in musical settings entail dynamic exchanges of control and power, and framings of AI-agents’ roles by human performers. We probe these framings and power-control exchanges through qualitative thematic lenses, drawing from post-phenomenology, matters of fact and concern and feminist science and technology studies. We contribute with a novel interdisciplinary analytical method as a tool for developers and designers of AI systems to help visibilise and examine the implicit, the wider connections and entangled filaments in Human–AI musical interactions.

In this paper, we use critical and agonistic modes of inquiry to analyse and critique a specific application of AI to music practice. It records a structured interdisciplinary dialogue between 1) a musicologist and social scientist and 2) an engineer in music and computer science, focusing on folk-rnn and Irish Traditional Music (ITM) as a case study. We debate the role of data ethics in AI music applications, the dynamics of inclusion and exclusion, and the nature of embedded value systems and power asymmetries inherent in applying AI to music. We discuss how identifying the value of AI music applications is critical for ensuring research efforts make musical contributions along with academic and technical ones. Overall, this agonistic dialogue exemplifies how questions of right and wrong — the core of ethics — can be examined as AI is applied more and more to music practice.

As music generated using artificial intelligence (AI music) becomes more prevalent — originating not only from individuals but also commercial services — the need to study it and its impacts becomes important. How can this material and its sources be meaningfully studied and critically engaged with, especially considering the unprecedented scales possible with generative AI? The paper begins to answer this question by considering AI music along seven aspects: 1) the company providing an AI music service; 2) its founders and employees; 3) the use of the service; 4) the users; 5) the algorithms; 6) the music; and 7) the sustainability. We make our discussion more concrete by considering the contemporary AI music service Boomy. While our investigations are preliminary and focused on a single AI music service, we argue that they open several interesting avenues of exploration for many disciplines and their intersections to help prepare for the coming flood of AI music. This paper asks many more questions than it answers, which is a feature (not a bug) of it advocating for a new domain of study: AI Music Studies.

Feverish narratives of artistic AI-revolution obscure the fact that empirical documentation of the actual impacts of artificial intelligence on artistic practices is still sparse. This paper focuses on the frictions of working with AI artistically. Based on interviews with 20 AI-artists, we 1) demonstrate that frictions experienced with the technological elements of the work processes with AI are inseparably intertwined with the artists’ socio-material realities and the inherent asymmetries of access, and 2) show how frictional ambivalence and unpredictability in artistic interactions with AI tools function both as restrictive and productive elements of the art-making processes, presenting opportunities to reframe the core notions of artistic agency, authorship, and the ontology of art.We discuss these findings in the context of HCI and critical data studies and provide three invitations for designing with and for frictions. Our empirical work contributes to a deeper understanding of the emerging community of AI-artists and invites new mindful perspectives for the design and development of Creative AI applications.

The environmental footprint of Generative AI and other Deep Learning (DL) technologies is increasing. To understand the scale of the problem and to identify solutions for avoiding excessive energy use in DL research at communities such as ISMIR, more knowledge is needed of the current energy cost of the undertaken research. In this paper, we provide a scoping inquiry of how the ISMIR research concerning automatic music generation (AMG) and computing-heavy music analysis currently discloses information related to environmental impact. We present a study based on two corpora that document 1) ISMIR papers published in the years 2017–2023 that introduce an AMG model, and 2) ISMIR papers from the years 2022–2023 that propose music analysis models and include heavy computations with GPUs. Our study demonstrates a lack of transparency in model training documentation. It provides the first estimates of energy consumption related to model training at ISMIR, as a baseline for making more systematic estimates about the energy footprint of the ISMIR conference in relation to other machine learning events. Furthermore, we map the geographical distribution of generative model contributions and discuss the corporate role in the funding and model choices in this body of work.