Activating language model to steer narrative flow through typhoon storm

Activation Steering the Nan-Shan in Typhoon

Computation and LanguageArtificial IntelligenceMachine LearningSound

Summary

This work uses a powerful text-generating AI to rewrite the story of a ship sailing into a typhoon, based on a classic novella. The AI takes the original sentences and keeps writing more text from its own output, sometimes becoming confusing or going off-topic. The authors introduce a new way to guide the AI's writing style and pacing called activation steering, which helps control how the story and side ideas unfold over time. This shows how AI can be somewhat controllable but still hard to fully understand or predict.

What this means in practice

  • For interactive narrative designers: Use activation steering to control long-form AI-generated story arcs with variable pacing and thematic digressions.
  • For digital artists: Create audiovisual installations with AI-generated text that evolves unpredictably yet remains partially guided by creative intent.

Authors

Megan Wei, Damon Pham, John Cayley, Ellie Pavlick

Abstract

Typhoon is an audiovisual installation based on Joseph Conrad's novella, in which Captain MacWhirr steers the steamer Nan-Shan directly into and through a catastrophic storm. The novella is a study of what happens when the scale of a phenomenon exceeds the practical and conceptual frameworks available to comprehend it and of the paralysis that sometimes follows a demand for action in such overwhelming circumstances. We chose the text of Typhoon to reflect on human culture's current AI moment. We prompt a large language model, seeded with Conrad's original language, to feed endlessly on its own output, and explore when and how the text becomes increasingly alien and unwieldy as time progresses, including deviations from the narrative, digressions into the model's own instructions and thinking, and long periods of incoherent rumination. We employ methods from activation steering, including a novel adaptive steering method we introduce, to control the rhythm of the narrative and the model's digressions over long stretches of generation. The simultaneous efficacy and unpredictability of this steering showcases the tension between the perceived interpretability of AI, and the reality of a system that remains outside our comprehension and control.