Sarah Fdili Alaoui is a professor at the Creative Computing Institute, University of the Arts London, and a choreographer, dancer and Laban Movement Analyst. Her public record matters to embodied AI because it offers a sustained alternative to technology-first dance research: movement knowledge, choreographic process and the politics of collaboration shape what is built and how it is evaluated. This is a source-verified profile, not an interview.

Context: one practice across studio, stage and research lab

Fdili Alaoui's official biography describes work at the intersection of human–computer interaction and dance, with computation integrated into embodied creative practice and choreography. It also records a PhD in Human–Computer Interaction and Art & Science from Université Paris-Sud 11, conducted at IRCAM–Centre Pompidou and LIMSI-CNRS, following training in applied mathematics and computer science. Before joining UAL, she was an associate professor at LISN, Université Paris-Saclay, and a member of the Ex)situ INRIA team (Fdili Alaoui, n.d.).

That combination can be narrated too easily as a bridge between two separate worlds: technical analysis on one side and dance on the other. Her documented work suggests a stronger proposition. Dance is not merely application material for an existing system. It supplies methods of attention, composition, documentation and critique that can reorganise interaction design.

Public record: movement quality as an interface problem

Early work on movement qualities made this orientation concrete. “Movement Qualities as Interaction Modality,” with Baptiste Caramiaux, Marcos Serrano and Frédéric Bevilacqua, treated qualitative movement descriptions as material for interaction design (Fdili Alaoui et al., 2012). The later ACM Transactions on Interactive Intelligent Systems article “Interactive Visuals as Metaphors for Dance Movement Qualities” examined how physical-model visuals could respond to qualities of dance movement (Fdili Alaoui et al., 2015).

These projects did not simply map a hand coordinate to a graphic coordinate. They addressed the difficult middle layer between measurable motion and lived, culturally trained description. A movement-quality term is neither a raw sensor channel nor a universal label. It is an interpretive construct. Designing with it requires decisions about who names a quality, how examples are selected and what a visual response makes perceptible.

That question remains central as AI systems learn motion embeddings and text-conditioned generation. The Lab's explanation of movement vocabulary describes the same translation pressure in current models; Fdili Alaoui's work shows that it predates the present generative-AI cycle.

Documented work: from tools for choreography to dance-led research

In “How Do Choreographers Craft Dance?”, Marianela Ciolfi Felice, Fdili Alaoui and Wendy Mackay studied choreographic work to frame technology as a partner rather than an automated replacement (Ciolfi Felice et al., 2016). “Making an Interactive Dance Piece” later used the making of a performance to analyse tensions that appear when technology enters artistic practice (Fdili Alaoui, 2019). The research contribution came through the artistic process, including its breakdowns and negotiations, rather than only through a final system demonstration.

Her 2023 habilitation thesis is titled Dance-led Research. The phrase is exacting. “Dance-led” does not mean that a project happens to include dancers; it means dance can lead the formation of research questions and methods. That stance is important in a field where a ready-made model may determine the available gestures, labels and measures before a collaborator enters the room.

Recent work extends the argument into generative systems. “Designing Movement Generation Models in Collaboration With Voguing and Dancehall Dancers,” published at CHI 2026, centres collaboration with practitioners of two movement forms whose histories and aesthetic criteria cannot be reduced to a generic motion dataset (Cheddin et al., 2026). The publication record therefore connects an early concern with movement quality to a current concern with who participates in model design.

A 2026 return to choreography as intimate, situated work

Fdili Alaoui's 2026 paper “An Autoethnography of Choreographing the Dance Performance ‘Vintage’” documents her own choreographic process and use of a web application for recording dances as annotated audiovisual clips. The abstract characterises choreography as intimate, relational, situated and emergent, and argues for technology that can remain in the background of the creative process (Fdili Alaoui, 2026).

This is a consequential counterweight to AI interfaces that demand continuous prompting, classification or optimisation. A tool can be computationally powerful yet choreographically intrusive. Background technology does not mean invisible politics or unexamined automation; it means the tool's demand for attention is designed around the practice rather than the reverse.

The same paper also demonstrates why autoethnography matters. First-person documentation can reveal the changing significance of a tool across rehearsal, memory and relationship. It does not replace external evaluation, but it can make questions visible that benchmark scores cannot pose.

Critical engagement: translation always changes what it carries

The strength of Fdili Alaoui's programme is also the location of its hardest problem. Movement qualities, choreographic annotations and co-designed generative models must still be translated into computational form. Translation creates exclusions. A label may stabilise what dancers use fluidly; an interface may privilege the collaborator most comfortable verbalising; a model may absorb examples while losing the social conditions that made them meaningful.

Participatory practice does not automatically resolve power. Researchers still control budgets, publication, system architecture and often the terms of data retention. Collaboration must therefore be evaluated through traceable decisions: who can refuse capture, who defines a successful generation, who can remove data and who receives credit.

The public record is valuable precisely because it does not offer a frictionless synthesis. “Dance-led” research asks technology to remain answerable to situations where ambiguity, intimacy and cultural history are not noise. That can make a project slower and its findings less easily generalised. It can also prevent the system from mistaking convenience for validity.

Field significance: changing the order of expertise

Embodied AI often begins with a technical capacity—pose estimation, a motion latent space, a generator—and then searches for a use. Fdili Alaoui's work reverses the order. It begins with how people move, make, learn and relate, then asks what computational intervention is warranted.

This affects evaluation. A motion generator cannot be assessed only by distributional similarity if a dancer needs selective revision, cultural legibility or space for surprise. An interactive visual cannot be assessed only by low latency if its metaphor misreads the movement. A documentation tool cannot be assessed only by retrieval speed if it interrupts the intimacy of remembering together.

For somatic AI, her significance lies less in a single algorithm than in a durable research arrangement: technical knowledge, artistic practice and first-person experience correct one another. That arrangement offers a practical basis for the Lab's ongoing questions about human–AI choreographic authorship and culturally situated movement data.

It also clarifies what “human-centred” should mean. The human is not an abstract user placed at the end of a pipeline. The moving person is a knowledge-holder whose practice can alter the categories, measures and purpose of the system itself.

References

  • Cheddin, L., Françoise, J., Caramiaux, B., & Fdili Alaoui, S. (2026). Designing movement generation models in collaboration with voguing and dancehall dancers. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3772318.3791515
  • Ciolfi Felice, M., Fdili Alaoui, S., & Mackay, W. E. (2016). How do choreographers craft dance? Designing for a choreographer-technology partnership. In Proceedings of the 3rd International Symposium on Movement and Computing. https://doi.org/10.1145/2948910.2948941
  • Fdili Alaoui, S. (2019). Making an interactive dance piece: Tensions in integrating technology in art. In Proceedings of the 2019 ACM Designing Interactive Systems Conference. https://doi.org/10.1145/3322276.3322289
  • Fdili Alaoui, S. (2026). An autoethnography of choreographing the dance performance “Vintage.” In Proceedings of the 2026 Conference on Creativity and Cognition (pp. 772–784). https://doi.org/10.1145/3803784.3807540
  • Fdili Alaoui, S. (n.d.). Sarah Fdili Alaoui: Researcher in human–computer interaction, choreographer, dancer, Laban Movement Analyst. Retrieved November 1, 2026, from https://saralaoui.com/
  • Fdili Alaoui, S., Bevilacqua, F., & Jacquemin, C. (2015). Interactive visuals as metaphors for dance movement qualities. ACM Transactions on Interactive Intelligent Systems, 5(3), Article 13. https://doi.org/10.1145/2738219
  • Fdili Alaoui, S., Caramiaux, B., Serrano, M., & Bevilacqua, F. (2012). Movement qualities as interaction modality. In Proceedings of Designing Interactive Systems. https://doi.org/10.1145/2317956.2318001