Unique Research · Arts · January 10, 2025
Stanford’s Michael Rau is reshaping theater for the digital age. As assistant professor of directing and dramaturgy in the Theater and Performance Studies (TAPS) program in the School of Humanities and Sciences, Rau has long explored how theater can reflect the complexity of modern life. With a seed grant from the Stanford Institute for Human-Centered Artificial Intelligence (HAI), he is now pushing that mission further: bringing AI into live performance.
His work spans three approaches:
- Image generation: An AI image generator takes a live video feed of the performance and, guided by prompts, modifies the actors’ images. The AI-adjusted images are projected onto the stage in sync with the live show — a stream that can read as dream sequences, alternate stories, or other dimensions.
- LLM live scripts: Audience suggestions are fed into a large language model that generates a script. Text-to-speech converts lines into audio; each character’s lines are split and sent to actors wearing earphones, who speak the AI-written dialogue. Actors need not hold scripts or stare at monitors — they can react in the moment to lines arriving in their ears.
- Pose analysis: Built for scholars, this tool uses AI to create a 3D body-position image of each actor for every frame of a recorded performance. Developed with Peter Broadwell, Simon Wiles, and Vijoy Abraham of Stanford Libraries’ Center for Interdisciplinary Digital Research, it could help film scholars study staging or political scientists analyze gestures in speeches.
Below, Rau describes how these projects spark debate about how technology can expand storytelling while keeping theater’s human core.
How did you get interested in theater and AI?
I firmly believe good theater reflects how we live now. Today our experience is heavily shaped by technology, yet you rarely see that reality on stage — watching someone use a device is not interesting. I wanted to make theater differently: figure out how people use technology, and how to show that effectively on stage. From 2014 I mounted several shows that folded technology into performance. In 2019, after seeing emerging tech, I became interested in machine-learning algorithms.
How does AI elevate theater?
I want the intersection where humans are still making art, and AI either enhances it or adds interesting new dimensions. The tech I’m using lets us layer a story on top of the staging the audience sees, create unrehearsed, exciting improvisational scripts, and give scholars entirely different tools to assess movement and pose. In every case, humans remain the primary focus.
Many worry AI art will be shallow, derivative, without meaning or intent. Is this art — or a gimmick?
I’ve wrestled with that. Art’s definition has always been subjective. This form has existed for centuries and tried almost every approach, so it is hard to say what is a gimmick and what is not. I apply AI in deliberate, narrow ways so I don’t strip away what is interesting and expressive in human performance. What reassures me about AI art is that I always include a human-in-the-loop element.
What are the benefits of AI in theater?
Honestly, we don’t yet know what the benefits will be. Given Stanford’s research mission, I think we are better off exploring these technologies — unexpected, exciting things may emerge. The alternative is to ignore them, which could freeze the art form in a way I don’t think helps. I hope AI gives us more interesting creative options, different ways of working, and a layer of concreteness and quantitative support for ongoing scholarly work.
What are the potential downsides?
First, serious IP and copyright issues are baked into the models; they must be resolved fairly. Second, I worry about homogenizing effects of AI in creative fields — for example, how content curation narrowly customizes our entertainment. Third, I worry about losing human-centered creativity. Making art is one of the most exhilarating things you can do — if we hand too much responsibility to ML systems, artists will suffer. In my work I try to keep human art-making at the center.
Will AI change theater’s essence, or the relationships among writers, actors, and directors?
Technology has always changed theater. We went from candlelight to gaslight to electric spots; that alone reshaped makeup, costume, set, and stage design, and many thought it would destroy theater. We are again discovering new tech and figuring out what is useful and how theater adapts.
What I love about this form is that both can coexist: you can see Shakespeare at the Globe in London, or West End shows that use every new stage technology. You can preserve a style or method — and you can try new ways of performing.
Which AI project are you excited about?
I’m developing a theater project tentatively titled Hamlet.AI with playwright Michael Yates Crowley. It centers on a playwright and a director who clash over using AI on stage, opening a debate and struggle about the future of art. The piece will fold in the image and text systems I’m using now, plus other ML techniques still in development. It will be a rich mix of technology and my complicated feelings about generative AI in art.
— END —