Radio Play: Live Participatory Worldbuilding and Performance with GPT



  • Radio Play: Live Participatory Worldbuilding and Performance with GPT



  • Presented by Robert Twomey & Ash Eliza Smith

    With recent advancements in machine learning, researchers have demonstrated remarkable achievements in natural language understanding and synthesis. This event introduces large language models (LLMs) through a participatory workshop culminating in a live radio show and internet broadcast. We explore aspects of liveness and serendipity; possibilities for human/non-human co-authorship; and relate emerging computational tools to human language and perception, ultimately considering how LLMs can be tools for creativity, co-creation, and live performance.

    The day-long workshop begins with an introduction to LLMs and generative text, then proceeds through a series of AI writers rooms, gaining experience with prompt authorship and interaction with GPT. We build towards human rehearsals of AI-generated text, culminating in a live radio show performance for both an in-person audience (30-40) and streamed to a large online audience. In this iteration, we build on previous workshops with the addition of live foley and real-time speech and voice synthesis. The final performance will be porous, inviting audience participation and involvement in shaping the development of the human/non-human radio drama.

    Together these elements extend our research into performative AI and narrative systems design. As artists and researchers we are interested in how we might co-author and co-perform with AI in a meaningful way, hinging on liveness and collective entropy. We are not just interested in what it means to collectively co-author (beyond the individual), but rather what the ingredients of liveness, audience participation, live sound scoring, and acting and performance bring to creative and performative AI.

    Our work ranges from real-time performance of audio and performance, speculative design, collective storytelling, live-action-role-play, and machine imagination.