Theatre creates a world in front of people who share a particular moment. In this part of my 2024 research into AI and visual culture, I ask how computational tools might change that world without losing the value of live presence. The possibilities concern writing, staging and access, as well as performances that deliberately make AI part of their subject.
In writing, AI could offer alternative scenes, dialogue or ways to develop a character. A playwright might use these proposals as material to accept, challenge or discard. More options would not automatically give a play greater depth. The work would still need a point of view, and choices about language, silence and conflict that belong to the production.
Casting invites a different distinction. Tools might help organise availability or review information supplied by performers, but they could not objectively determine the right actor from appearance or past roles. Unexpected casting can change the meaning of a production. A director's encounter with a performer should not be reduced to a score or a supposedly precise match.
For sets, costumes and sound, AI could help explore references and compare proposals before fabrication or rehearsal. Historical resemblance would need checking, and a convincing render would not establish that a set can be built or used safely. Lighting and projections might respond to rehearsed cues or live movement. Their contribution would be to support the scene, not merely demonstrate that the stage can react.
Rehearsal tools could assist scheduling and help a team revisit timing, movement and relationships on stage. Recordings, where performers agree to them, might make discussion more specific. An analysis would remain one interpretation rather than a verdict on the performance. Pauses, hesitation and departures from a pattern can be artistically significant.
Interactive performances could let an audience choose a route or influence part of a story. Virtual and augmented elements might extend a space beyond its physical scenery. Such participation would need clear boundaries and room for those who prefer to watch. A production could also foreground generated material or a synthetic performer, while making its construction part of the audience's understanding.
Speech processing, translation and surtitles could help more people follow a performance. They would need to respect timing, poetic language and the expressive qualities of a voice, rather than treating every accent as something to correct. Audience data might inform planning, but it would not reliably predict artistic success or replace the experience of testing a work with people.
I imagine AI expanding the theatre's working vocabulary while making its human choices more visible. A shared stock of images and stories could be reinterpreted for one room, one cast and one evening. The most interesting result would not be a technically perfect spectacle, but another way to make a live encounter matter.

AI & Visual Culture: 40. Virtual Reality ↗

