My response to Pelinski et al's at the AIMS26 “Beyond Ethics, for a Critical Interdisciplinary AI Music Pedagogy” track

I travelled to Nottingham this week for the AI Music Studies (AIMS) 2026 conference, where I gave a response to Pelinski et al.’s paper in the session “Beyond Ethics, for a Critical Interdisciplinary AI Music Pedagogy.” Below is the response as I delivered it at the conference.

I would like thank Teresa Pelinski and co-authors for a paper that is both timely and ambitious.

What I find most valuable about this piece, is that it argues that ethics, when taught as an add-on (or as an elective or optional module) within a technical curriculum like computer science, risks reinforcing techno-solutionism. Techno-solutionism is this belief that engineering or technology can solve all human and societal problems.

Instead, the paper argues that ethics as an elective, may be too narrow, and that students should learn to understand and frame technology relationally, in connection with social and cultural contexts. It proposes Music AI as an exemplar for building a pedagogy with a broader, interdisciplinary remit, capable of changing how technical problems are framed in the first place.

So, a pedagogy that shifts student’s mindset from asking questions such as How do we mitigate the harms of AI? (a more typically framed AI ethics question) and coming up with technical fixes such as fine-tuning specific models, and ‘‘debiasing’’ datasets, and making them more inclusive/representative and balanced.

And moving towards another mindset which asks a prior, more critical set of questions: how are AI systems already shaped by social assumptions, by institutions, by histories, by relations of power? What values and worldviews embedded in them as they are presented to us as apparently neutral tools? And whose interests do they end up serving?

And I think this is very valuable.

Now, if ethics-as-add-on is inadequate, what would a computing education curriculum look like if it framed technology and AI music as social, cultural, political, and relational from the outset?

The 8-week MusAI program led by the project “Prototyping Radically Interdisciplinary Music AI pedagogies”, which I had the opportunity to attend as an observer-participant, focused on AI music as an example application area.

And the program showed us that, the moment one starts dealing with musical genre, recommendation, listening, creativity, performance, taste, or musical value, one runs into questions that cannot be answered technically alone. Because we are dealing with historically sedimented categories, aesthetic judgments, institutional priorities, commercial pressures, and lived musical practices.

And I think the paper is absolutely right to insist that if students are not taught to perceive those dimensions, then they might be being trained to misrecognise the object they are working on.

Thinking across the MusAI programme as a whole, this paper reflects a pedagogy that has tried to move consistently across perspectives and scales. What I liked about the proposed themes and perspectives is that, together they build an argument: that AI music must be understood at once technically, aesthetically, socially, politically, historically, and institutionally.

Having had the benefit of experiencing some of this pedagogy at the seminar series at Queen Mary University of London, I would say that one of its achievements lies in not smoothing over disciplinary difference, but in enabling diffractive dialogues, making those differences audible and productive, by finding interference points. The moments of hesitation, disagreement, and translation — these were also the pedagogy of MusAI.

So my response, is that this paper matters because, it is not just asking for better training or proposing a curriculum. It is proposing a reorganisation of the field itself. So the paper leaves us with this important and ambitious challenge.

Perhaps this is where I would gently press it in a few points.

There was very positive student feedback indicating “that they especially appreciated the opportunities for interdisciplinary dialogue and deliberation”, but I was curious to read about where did the difficulties lie. For instance what were barriers to entry in the interdisciplinary discussion, what were instances of loss in translation, of friction, disengagement, and failure to understanding?

Another important point, addresses curriculum design and organisation.

Playing the devil’s advocate here, hybrid curricula with core modules plus interdisciplinary electives is built on the idea that students should first gain a solid grounding in a set of established disciplines, then learn to connect and apply them. This has benefits, such as stronger disciplinary depth and coherence, lower risk of superficiality. It is also considered a safer and more robust model, more manageable to deliver, with easier staffing and departmental ownership, and better accreditation and external validation.

Looking back to my 20-year-old-self computer science student, and later on, to my teaching experiences in computing education, that has been my experience.

A fully interdisciplinary curriculum starts from the assumption that many important problems do not fit neatly inside disciplines, so students learn through integration from the beginning, with problem-framing across domains, integrative thinking, and work in emerging or boundary-crossing fields.

Looking back to my second masters in management of creative and cultural industries, a joint program delivered by two departments, management and economics and the school of arts and creative industries – this was what I experienced.

A fully interdisciplinary curriculum can be more powerful, but I think it can only work well if the program is very carefully designed so that breadth does not replace depth and rigor, and ensures students develop mastery in methods, concepts, and real practice rather than assuming integration alone is enough.

So I would further press with the following questions:

What kind of graduate is the curriculum trying to produce? I would say critical computing practioners and well-informed citizens.

What forms of patience, translation, and even friction, have to be protected if different disciplines are genuinely to interface and change one another?

What would it take for this interdisciplinarity to be institutionalised? What kinds of curricula and learning outcomes, interdisciplinary interfaces and evaluation criteria, faculty profiles and core skills, departmental collaborations, funding structures, and professional incentives would be needed?

And if are we prepared to take the harder path of building institutions in which critical interdisciplinarity becomes part of computing education, what leverage points can we make use of for pursuing this?

I think we should take this paper seriously, take the challenge, and become prepared. I believe that these times of ‘disruption’ of the education system, where the meaning and value of academia and university are many times questioned, might also be times of opportunity for designing and deploying radical interdisciplinary approaches to computing pedagogy.