Maya Man’s AI dancers turn perfection into an endless rehearsal

In Maya Man’s imaginary dance competition, the costumes can remain convincing while the bodies come apart. The bitforms gallery catalogue describes legs bending impossibly and two dancers merging into one. StarQuest borrows the makeup, backstage rituals and anxious confessionals of competitive dance, then lets the machinery producing them show through.[1]

The V&A presented its London performance-lecture on 10 October 2026, combining Man’s software work, online material and choreography. The artist wrote and performed the piece, directed by Mackenzie Davenport. Its subject is the relationship between American competition dance and an online world in which presenting yourself can become a continuous, measurable activity.[2]

Man knows the first of those worlds from inside it. Speaking to Louis Jebb for Right Click Save, she recalled growing up as a competition dancer in suburban Pennsylvania and training for hours each day. Revisiting that experience helped her understand why posting online had so often felt like stepping onto another stage.[3]

In a separate Hypeart interview, she described making StarQuest as a shift from dancer to coach. The fictional coach’s demand for exceptional performance echoed a standard she still imposed on herself. Generating the performers gave her the power to direct a situation she had once inhabited.[4]

The performance changes after the clips are made

The underlying work debuted on Feral File in November 2025. It contains 111 AI-generated scenes, each eight seconds long, made with text-to-video and text-to-music models. They form a fictional television season inspired by Dance Moms, with custom software assembling the scenes into changing sequences.[1]

That distinction matters to Man. She told Jebb that StarQuest’s medium is custom software: a browser shuffles the clips while the work runs. The generative part of a viewing therefore includes their arrangement. It does not require the video model to invent another dancer each time the next scene appears.[3]

A fixed scene can take on another role when its neighbours change. Place a confession after a triumph and it can undercut the triumph; place it before the performance and it can become anticipation. Those are illustrative possibilities, but they explain why choosing the sequence is a creative act in its own right.

The artwork gives its maker two kinds of control: producing the scenes and deciding how they can meet. It also leaves room for surprise inside the second task. The footage is finished; the relationships between its parts remain in play.

A fictional show acquires its own fan culture

For the StarPower exhibition at bitforms, the world extended into short fan-style edits made with CapCut templates. These appeared on actual iPhones. Material from the central work became the sort of compressed, circulating fragment through which a television personality might reach another audience.[1]

Man explained to Jebb that the scale was deliberate. A large central display represented the primary show; the phones held its remixed, secondary life. The installation made the devices part of the sculpture, rather than treating every screen as an interchangeable container.[3]

The comparison is unusually apt for generative media. Making a persuasive image is one task. Giving that image a recurring place in someone’s attention is another. StarQuest builds both into its fictional world: the performance and the material that keeps the performance circulating.

Man told Hypeart that watching the invented characters had affected her emotionally even though she knew they did not exist. That reaction unsettled her. The scene had no real dancer behind it, yet the pressure it carried was familiar to the person who had made it.[4]

Report an error or suggest a correction