Exploring the Collaborative Co-Creation Process with AI: A Case Study in Novice Music Production

Yue Fu, Michele Newman, Lewis Going, Qiuzi Feng, Jin Ha Lee · 2025

Read Exploring the Collaborative Co-Creation Process with AI: A Case Study in Novice Music Production by Yue Fu, Michele Newman, Lewis Going, Qiuzi Feng, Jin Ha Lee (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-15276.txt.

Introduction

Yue Fu and co-authors study novices making music with AI in a 2025 paper. Nine undergraduates in one class made three songs with AI tools in 10 weeks. Songs went from first spark to a Spotify release. AI helped with tunes, words, cover art, and the release path. Logs and talks caught how the work flow bent.

AI sped the spark stage and shrank the prep stage of old craft. Pick and proof of raw outputs then grew hard for new hands. A new stage rose: collage and refine, where parts fuse to wholes. Roles in the group shifted with the tools as well. Two models of the co-work close the paper.

Catalog record

CREATORYue Fu, Michele Newman, Lewis Going, Qiuzi Feng, Jin Ha Lee
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE80 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-15276

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Exploring the Collaborative Co-Creation Process with AI: A Case Study in Novice Music Production Yue Fu, Michele Newman, Lewis Going, Qiuzi Feng, Jin Ha Lee Abstract . Artificial intelligence is reshaping creative domains, yet its co-creative processes, especially in group settings with novice users, remain under explored. To bridge this gap, we conducted a case study in a college-level course where nine undergraduate students were tasked with creating three original music tracks using AI tools over 10 weeks. The study spanned the entire creative journey from ideation to releasing these songs on Spotify. Participants leveraged AI for music and lyric production, cover art, and distribution. Our findings highlight how AI transforms creative workflows: accelerating ideation but compressing the traditional preparation stage, and requiring novices to navigate a challenging idea selection and validation phase. We also identified a new “collaging and refinement” stage, where participants creatively combined diverse AI-generated outputs into cohesive works. Furthermore, AI influenced group social dynamics and role division among human creators. Based on these insights, we propose the Human-AI Co-Creation Stage Model and the Human-AI Agency Model, offering new perspectives on collaborative co-creation with AI. CCS: Human-centered computing Empirical studies in HCI 1. Introduction Artificial intelligence has afforded a new era of creativity, where humans and machines collaborate to produce novel and innovative works. Human-AI co-creation has emerged as a growing field of interest, capturing the attention of researchers and practitioners alike. The goal is to leverage the strengths of both humans and AI to produce creative outcomes that go beyond what either could achieve alone ( Wingström et al., 2024 ) . Such synergy has found applications across various domains, including writing ( Osone et al., 2021 ) , fashion design ( Deng et al., 2024 ) , architecture ( Tan and Luhrs, 2024 ) , marketing and advertising ( Islam et al., 2024 ) , visual arts ( Zhou and Lee, 2024 ) , and more. Figure 1 . The Human-AI Co-Creation Stage Model. The diagram shows the creative process with AI support along a temporal dimension. As discussed in Section 4.1.2 , introducing AI often shortens or bypasses Stage 2 (Preparation)—shown here with dashed boundaries. Although AI is praised for facilitating idea generation in Stage 3, it can also quickly lead to convergent stages in Stages 4 and 5, potentially linearize the creative process (see Section 5.3 ). Additionally, we identify a new Stage 5 (Collaging, Refining, and Integration)—marked by an asterisk—where creators must merge AI-generated and human-created elements. Should this piece-together process fail, users revert to Stage 3 to regenerate compatible outputs. Involving AI also facilitate achieving tangible results, fosters a “progress loop” ( Amabile and Pratt, 2016 ) , giving creators a sense of accomplishment and prompting them to revisit Stage 1 for subsequent creation cycles. Finally, AI can be viewed as an environmental factor that supports user autonomy, competence, or task involvement throughout the stages, with the exception of the end outcome (see Section 5.1.1 ). In music production, AI-powered tools have become increasingly popular, supporting both expert musicians ( Bryan-Kinns et al., 2024 ; Banar et al., 2023 ) and novices ( Louie et al., 2020a ; Kobayashi et al., 2024a ) in the creative process. A plethora of AI music generation and production tools powered by foundational models ( Liu et al., 2023 ) have been developed, aiming to facilitate music creation by increasing efficiency, enhancing self-expression, and supporting idea generation. AI music tools can compose melodies ( Ju et al., 2021 ) , harmonize chords ( Van Kranenburg et al., 2023 ; Tsushima et al., 2017 ) , transcribe ( Rigaud and Radenen, 2016 ) , improvise ( Madaghiele et al., 2021 ) , suggest lyrics ( Wang and Lin, 2024 ) , and assist in mixing and mastering tracks ( Shtern et al., 2018 ) . For novice users, AI lowers barriers to entry, enabling individuals without formal training to engage in music production and express their creativity. Platforms like Suno ( sun, 2025 ) , Google’s Magenta ( mag, 2025 ) , and OpenAI’s MuseNet ( Mus, 2025 ) advertised they can democratize music creation by providing users with accessible tools that augment their creative capabilities. Despite the growing prevalence of AI in creative fields, there is a notable gap in understanding the detailed processes of how creators collaboratively co-create with AI over extended periods. Much of the existing research has been conducted in laboratory settings or through online studies that capture only a snapshot of the creative process—often within a single session lasting less than an hour, using toy creative problems, and does not lead to final usable creative output. However, real-world creative artifact creation is typically a prolonged endeavor that involves various creative working stages and unfolds over weeks or months. In fields like architecture, design projects can span years, and in music production, though varying by genre and scope, it often takes weeks for a professional production team to produce a track from the initial concept to the final release. Moreover, for much of creative processes, creation is a collaborative endeavor ( Sawyer, 2017 ) . In disciplines such as design, tools like Figma ( Figma, 2025 ) and Miro ( mir, 2025 ) have revolutionized how teams work together, providing ways for people to collaboratively brainstorm, vote, and comment. Similarly, in music production, collaboration is the backbone of the creative process. It usually involves a team of individuals fulfilling various functions, including lyricists, composers, mixing and mastering engineers, and more. Each role contributes specialized expertise, and the interplay among team members is crucial to the success of the final outcome. Using novice music production as a case study, our study aims to explore how AI tools influence collaborative dynamics and shape the stages of co-creation over an extended period. Music production represents an instructive example because it spans multiple creative roles (composition, arrangement, lyric writing, mixing/mastering) and often requires iterations to achieve a final outcome. Understanding both the collaborative dynamics and process of novice music production is essential for designing AI systems that effectively support collaborative co-creation activities between humans and AI. Specifically, we investigate three research questions: • RQ1: What are the temporal stages of the co-creative process when novice users collaborate with AI in music production? • RQ2: What relationships and roles do humans and AI assume in the collaborative music co-creation process? • RQ3: How do users perceive AI in this process? To address these questions, our case study explores how novice music creators work as a team to co-create music with AI over a period of 10 weeks. The setting of our study is an undergraduate capstone class at our institution, where students with no or limited previous music production experience used AI tools individually and collaboratively to create music pieces. They explored various AI technologies for different stages of music creation, including composition, arrangement, lyric writing, mixing and mastering, etc. Our research team reviewed each team’s artifacts, including course submission materials, process logs, published songs, group websites, and final presentations. We then conducted in-depth interviews with nine participants to understand their experiences, the social and technical challenges they faced, and how AI shaped their creative and collaborative

Read the complete text