CoChat

Reflection Across AI-based Music Composition

Open access

The DOI matches its Crossref record; 72 of the 72 references it cites were found in a registry or an index; 9 of the 101 sources its claims cite were checked for support.

Abstract

Reflection is fundamental to creative practice. However, the plurality of ways in which people reflect when using AI Generated Content (AIGC) is underexplored. This paper takes AI-based music composition as a case study to explore how artist-researcher composers reflected when integrating AIGC into their music composition process. The AI tools explored range from Markov Chains for music generation to Variational Auto-Encoders for modifying timbre. We used a novel method where our composers would pause and reflect back on screenshots of their composing after every hour, using this documentation to write first-person accounts showcasing their subjective viewpoints on their experience. We triangulate the first-person accounts with interviews and questionnaire measures to contribute descriptions on how the composers reflected. For example, we found that many composers reflect on future directions in which to take their music whilst curating AIGC. Our findings contribute to supporting future explorations on reflection in creative HCI contexts.

Overview

CoChat Summary

TL;DR

This paper explores how six artist-researcher composers reflected while integrating different AI music tools into their composition practice, using hourly reflection boards, interviews, RiCE questionnaire measures, and first-person accounts. It contributes six first-person accounts and thematic descriptions of reflection, including reflection on future directions while curating AIGC.

Key findings

  • Thematic Analysis produced six themes: Reflection on Past Instincts; Reflection on Direction and Surprises; Reflection for AI; Reflection on Feelings; Reflection on Influences; and Reflection on Technical Challenges.
  • RiCE participant averages (n=7) were reflection-on-process 8.1, reflection-on-self 5.8, reflection-through-experimentation 6.6, and total RiCE 6.8; the open-source all-CST comparison was 7.4, 6.1, 6.9, and 6.8 respectively, while the DAW subset was 7.6, 5.7, 7.1, and 6.8.
  • Reflection-on-process scores were all high, with none falling below four; P1, P3, and P4 peaked in Sessions 1 and 3, while P2, P5, and P6 generally decreased after Session 2. Reflection-on-self peaked in Sessions 2 and 4 for P1–P5, but P6 and P7 peaked in Session 3; reflection-through-experimentation tentatively declined over time.
  • Composers reflected on future directions while curating AIGC, and reflection-on-surprise arose from both AI and non-AI workflow moments, such as P5’s unexpected cowbell sample.

Methods

  • Seven composers were recruited (two collaborated as a band, yielding six compositions); all were PhD students aged 18+, had integrated AI into their music making, and used self-selected AI tools. They completed four one-hour composition sessions remotely, each producing a minimum one-minute composition.
  • Data collection after each hour included the RiCE v2 questionnaire, Miro reflection boards with 6–10 chronological screenshots and post-it reflections, and semi-structured interviews; after all sessions, participants wrote 800–1000 word first-person accounts.
  • Analysis used inductive Thematic Analysis on collated reflection boards, interview transcripts, and first-person accounts, triangulated with questionnaire measures; no statistical analysis was conducted due to the small sample size.

Limitations

  • Stated in the paper: participant backgrounds, tools, compositional techniques, and genres varied widely, preventing statistical analysis or generalisation; the authors acknowledge conflation between tool types and call for more controlled A/B tests.
  • Stated in the paper: the structured hourly pauses meant participants did not self-select reflection moments, and the method required retrospective synthesis into first-person accounts rather than capturing immediate diary-style thoughts; interviews were also a contingency in case accounts were not completed.
  • Evident from the text: the sample is only seven artist-researchers with high musical sophistication (all 75%+) and self-reflection scores (all over 70%), so findings may not represent broader composer populations.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1145/3635636.3656185

Cites

72 of 72 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

8 Valid · 2 Not valid · 51 Not checked · 99 Own results and interpretations

46 more claims

Create a free account to see all 160 claims and to verify the 92 references still open.

See all claims — free account
  • Probably not a substantial claim. 88
  • It may state two separate claims as one. 48
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

3 of 3

Citations

Cites

Show all 72 references

Research works printed but not linked: 25

  1. Berker Banar and Simon Colton. 2021. Generating Music with Extreme Passages using GPT-2. Evo* 2021 Late Breaking Abstracts (2021), 31. https://arxiv.org/pdf/2106.11804.pdf#page=34 · arXiv
  2. V. Braun and V. Clarke. 2013. Successful Qualitative Research: A Practical Guide for Beginners. SAGE Publications.
  3. Jean-Pierre Briot Gaëtan Hadjeres and François-David Pachet. 2017. Deep Learning Techniques for Music Generation – A Survey. https://doi.org/10.48550/arxiv.1709.01620 · DOI
  4. Nick Bryan-Kinns, Berker Banar, Corey Ford, Courtney N. Reed, Yixiao Zhang, Simon Colton, and Jack Armitage. 2021. Exploring XAI for the Arts: Explaining Latent Space in Generative Music. In 1st Workshop on eXplainable AI Approaches for Debugging and Diagnosis (XAI4Debugging@NeurIPS2021). https://xai4debugging.github.io/files/papers/exploring_xai_for_the_arts_exp.pdf
  5. Antoine Caillon and Philippe Esling. 2021. RAVE: A Variational Autoencoder for Fast and High-quality Neural Audio Synthesis. arXiv preprint (2021). https://arxiv.org/abs/2111.05011 · arXiv
  6. Baptiste Caramiaux Fabien Lotte Joost Geurts Giuseppe Amato Malte Behrmann Frédéric Bimbot Fabrizio Falchi Ander Garcia Jaume Gibert Guillaume Gravier Hadmut Holken Hartmut Koenitz Sylvain Lefebvre Antoine Liutkus Andrew Perkis Rafael Redondo Enrico Turrin Thierry Viéville and Emmanuel Vincent. 2019. AI in the Media and Creative Industries. Research Report. New European Media (NEM). 1–35 pages. https://inria.hal.science/hal-02125504
  7. John Chowning. 1977. Stria. Commissioned by IRCAM (Paris) for the Institute’s first major concert series: Perspectives of the 20th Century.
  8. Zihang Dai Zhilin Yang Yiming Yang Jaime Carbonell Quoc V. Le and Ruslan Salakhutdinov. 2019. Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context. arxiv:1901.02860 [cs.LG] · arXiv
  9. Carolyn Ellis, Tony E. Adams, and Arthur P. Bochner. 2011. Autoethnography: An Overview. Historical Social Research / Historische Sozialforschung 36, 4 (138) (2011), 273–290. http://www.jstor.org/stable/23032294
  10. Sara Fdili Alaoui. 2023. Dance-Led Research. Computer Science. Université Paris Saclay (COMUE). https://inria.hal.science/tel-04059520/
  11. Corey Ford and Nick Bryan-Kinns. 2022. Speculating on Reflection and People’s Music Co-Creation with AI. In Generative AI and HCI Workshop at CHI 2022. https://qmro.qmul.ac.uk/xmlui/handle/123456789/80144
  12. Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2014. Generative Adversarial Nets. Advances in Neural Information Processing Systems 27 (2014).
  13. Théo Jourdan and Baptiste Caramiaux. 2023. Machine Learning for Musical Expression: A Systematic Literature Review. In Proceedings of the International Conference on New Interfaces for Musical Expression. Mexico City, Mexico. https://hal.science/hal-04075492
  14. Diederik P Kingma and Max Welling. 2013. Auto-Encoding Variational Bayes. https://doi.org/10.48550/ARXIV.1312.6114 · DOI · Paper page
  15. Makayla Lewis. 2023. AIxArtist: A First-person Tale of Interacting with Artificial Intelligence to Escape Creative Block. In Proceedings of the 1st International Workshop on Explainable AI for the Arts (XAIxArts), ACM Creativity and Cognition (C&C) 2023. https://arxiv.org/abs/2308.11424 · arXiv
  16. Jackson Loth, Pedro Sarmento, CJ Carr, Zack Zukowski, and Mathieu Barthet. 2023. ProgGP: From GuitarPro Tablature Neural Generation to Progressive Metal Production. In Proceedings of the 16th International Symposium on Computer Music Multidisciplinary Research. https://arxiv.org/abs/2307.05328 · arXiv
  17. Michael Muller, Justin D Weisz, and Werner Geyer. 2020. Mixed Initiative Generative AI Interfaces: An Analytic Framework for Generative AI Applications. In Proceedings of the ‘The Future of Co-Creative Systems-A Workshop on Human-ComputerCo-Creativity” at the 11th International Conference on Computational Creativity (ICCC 2020).
  18. Ashley Noel-Hirst and Nick Bryan-Kinns. 2023. An Autoethnographic Exploration of XAI in Algorithmic Composition. In The 1st International Workshop on Explainable AI for the Arts (XAIxArts), ACM Creativity and Cognition (C&C) 2023. https://arxiv.org/abs/2308.06089 · arXiv
  19. Pauline Oliveros. 2005. Deep Listening: A Composer’s Sound Practice. IUniverse.
  20. Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, and others. 2019. Language Models are Unsupervised Multitask Learners. OpenAI Blog 1, 8 (2019), 9.
  21. P. Sarmento, A. Kumar, CJ Carr, Z. Zukowski, M. Barthet, and Y. Yang. 2021. DadaGP: A Dataset of Tokenized GuitarPro Songs for Sequence Models. In Proceedings of the 22nd International Society for Music Information Retrieval Conference (Online). https://arxiv.org/abs/2107.14653 · arXiv
  22. Pedro Sarmento, Adarsh Kumar, Yu-Hua Chen, CJ Carr, Zack Zukowski, and Mathieu Barthet. 2023. GTR-CTRL: Instrument and Genre Conditioning for Guitar-focused Music Generation with Transformers. In International Conference on Computational Intelligence in Music, Sound, Art and Design (Part of EvoStar). Springer Nature Switzerland Cham, 260–275.
  23. Bob Sturm. 2022. Generative AI Helps One Express Things for Which They May Not Have Expressions (Yet). In Workshop on Generative AI and HCI at the CHI Conference on Human Factors in Computing Systems 2022. https://kth.diva-portal.org/smash/get/diva2:1757906/FULLTEXT01.pdf
  24. Elizabeth Wilson Shawn Lawson Alex McLean and Jeremy Stewart. 2021. Autonomous Creation of Musical Pattern from Types and Models in Live Coding. In xCoAx 2021 9th Conference on Computation Communication Aesthetics & X. 76–93. https://qmro.qmul.ac.uk/xmlui/handle/123456789/73475
  25. K. Worrall, Z. Yin, and T. Collins. 2022. Comparative Evaluations in the Wild: Systems for the Expressive Rendering of Music. IEEE Transactions on Artificial Intelligence (2022). (Under review).

Cited by

See all 23

Metadata

Authors
Ford, C., Noel-Hirst, A., Cardinale, S., Loth, J., Sarmento, P., Wilson, E., Wolstanholme, L., Worrall, K., & Bryan-Kinns, N.
Affiliations
  • Queen Mary University of London (GB)
  • Creative Computing Institute, University of the Arts London, United Kingdom
Venue
Creativity and Cognition
Publisher
ACM
Published
2024-06-23
Pages
398-412
OpenAlex
W4399917657
Cited by
22
References
70
Citations by year
2026: 7 · 2025: 13 · 2024: 2
FWCI
4.53
Type
conference paper
Language
en
Locations
  • Creativity and Cognition (publishedVersion)
  • University of the Arts London Research Online (University of the Arts London) (acceptedVersion, cc-by)
  • Queen Mary Research Online (Queen Mary University of London) (submittedVersion, cc-by)
  • University of the Arts London Research Online (University of the Arts London) (submittedVersion)
  • Edge Hill University Research Information Repository (Edge Hill University) (submittedVersion, other-oa)
Funders
UK Research and Innovation (EP/S022694/1,EP/L01632X/1,EP/S022325/1)
Topics
  • Social Sciences › Arts and Humanities › Music › Music Technology and Sound Studies
  • Social Sciences › Arts and Humanities › Visual Arts and Performing Arts › Art, Technology, and Culture
  • Physical Sciences › Computer Science › Human-Computer Interaction › Innovative Human-Technology Interaction
Keywords
reflection, AI-generated content, creative practice, personal account, variational autoencoder, timbre modification, human-AI collaboration

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (qwen3.8-omni-flash and gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Reflection Across AI-based Music Composition. https://app.cochat.ai/papers/10.1145/3635636.3656185
The paper
Ford, C., Noel-Hirst, A., Cardinale, S., Loth, J., Sarmento, P., Wilson, E., Wolstanholme, L., Worrall, K., & Bryan-Kinns, N. (2024). Reflection Across AI-based Music Composition. Creativity and Cognition. https://doi.org/10.1145/3635636.3656185

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.