At AMIW 2026, the following distuingished researchers will give invited talks. Don't miss out!
Artists, Music AI and Training Data Attribution
Generative music AI lived for a long time in academic labs then entered the realworld quite suddenly this decade, in a form that aggressively scraped creative data to train generation tools. While a debate still rages about the application of copyright to AI training data, much work has been underway to create fair and artist friendly systems that respect the idea that training data should only be used with the consent of artists, in particular methods that seek to trace attribution from AI outputs back to AI training data sets. In my recent work I’ve been critical of such efforts and argue that they are ill defined and fail to convert principles of fairness into workable systems without seriously obfuscating the process and are easily gameable. I argue that they need careful oversight, and in recent work with Kathy Bowrey on policy design, argue that artist groups need to play a leading role in deciding what principles of fairness and credit apply in their specific creative niche. However, worse still, it is reasonable now to anticipate a world in which AI creativity is decoupled from training data, for example involving user reinforcement learning. The fight over attribution and training data may be surpassed by a new reality where that data no longer has any leverage. I consider what effects this has for creative policies, and for building an ethical research field.
Oliver (Ollie) Bown is a British researcher and musician based in Australia, where he is professor at the School of Art & Design at the University of New South Wales, Sydney, He is co-director of UNSW’s Creative Technologies Research Lab (CTRL), an interdisciplinary academic group comprising creative artists and musicians, designers, media theorists and cultural studies researchers who together research technology in its relation to creative work. Ollie’s research explores the nature of creativity in the arts, how technology is a site of artistic creative exploration, and how AI technologies are transforming creative fields in their ability to automate aspects of the creative process.
He has written over 100 academic articles in technology, arts and social science fields. His first book, Beyond the Creative Species: Making Machines that Make Art and Music, was published by MIT Press in 2021 immediately before the current boom in generative AI. Since 2022 he has been a researcher on the project Music & AI: Critical Interdisciplinary Studies funded by the European Research Council and bringing together international leaders in the social study of music and AI. He is currently also co-investigator on two Australian Research Council Discovery Projects: one on sound in public space; and the other on artists, AI and fairness. He is a regular commentator about the impacts of AI on the creative arts, and has contributed to debates at the World Intellectual Property Organisation, OECD, the Australian National AI Centre, Creative Australia and SXSW.
He is also an electronic musician best known for his work in the UK duo Icarus, since 1997, and Australian ensemble Tangents, since 2010.
Has it all been done? Music AI research in the age of LLMs
Foundation models can now generate convincing music from a text prompt, raising a provocative question for researchers: has it all been done? In this talk, I show that the most interesting problems in music AI are only now coming into focus. Drawing on recent work from the AMAAI Lab, including Mustango, Text2midi, Video2Music, MusicRL-style preference alignment, and SonicMaster for prompt-driven mastering, I map the frontiers that remain open: fine-grained controllability, aligning models with human musical preference, meaningful evaluation, multimodal and symbolic generation, AI tools for production and mixing, and explainability. I close with the responsible-AI challenges of datasets, copyright, and artist rights, and offer a roadmap of where researchers can still make outsized impact.
Dorien Herremans is an Associate Professor at the Singapore University of Technology and Design (SUTD), where she leads the Audio, Music, and AI (AMAAI) Lab. Her research sits at the intersection of machine learning and music, spanning controllable music generation (text-to-music, text-to-MIDI, and video-to-music), music understanding, AI-assisted music mastering, and more. Her lab has developed widely used open models and datasets such as Mustango, Text2midi, Video2Music, and SonicMaster. Before joining SUTD, she was a Marie Skłodowska-Curie Postdoctoral Fellow at the Centre for Digital Music at Queen Mary University of London. She was named to the Singapore 100 Women in Tech list in 2021 and was one of the top 30 finalists for the SAIL (Super AI Leader) Award at the World AI Conference in 2024.