📍Yokohama, Japan

On Sunday, April 29th, 2025, I attended the HCI Design Stories workshop, where HCI scholars who believed in the power of storytelling discussed its nuances and wondered about how to use stories as a research method. One key thread we discussed was how storytelling can make their subject matter accessible to story tellers and story listeners. The design fiction at the heart of the paper I’m presenting at CHI 2025 has, until now, been primarily about myself and my co-authors as storytellers (and during revisions, ourselves and reviewers as story editors).

The tale of Polly, a fictional emotional support AI, was written from the start as a research artifact, but also for me was an expression of existential dread. Surrounded by the seemingly indefatigable race to make generative AI as widely adopted for as many purposes as possible, I frequently find myself personally and professionally wishing I could somehow get ahead of it, even for a moment, and make everything just slow down. Storytelling about the future has always been a way to get ahead of ourselves, to make ourselves a little (justifiably and appropriately) spooked. Aside from any intellectual conversations that I hope are started by the specific examples and topics covered in this paper, I also believe that emotionality is right at the heart of what makes stories uniquely valuable as complements to other forms of research practice. Stories give us narratives that make complex topics sequential, characters to (not) relate to, and are, ideally, specific enough to make us feel/believe and ambiguous enough to make us think.

The format of a news article for the design fiction was chosen from a half a dozen other ideas in part because it echoed the current articles and headlines that (justifiably and appropriately) spook us: suicide-themed chatbots? sexual conversations with minors? Headlines, as usual, seem to tell the most troubling version of the story. But, at least from my perspective, the world of emotional support AI is most troubling not in headlines, but in the messes of trust, loneliness, exposure, and vulnerability that start to emerge when you when you dig into the stories we hear about chatbots in press and academia.

This paper is an attempt to start a comprehensive dissection of this mess; each time I talk about its discussion points with a new person, I find a new thread to pull at. Story telling (to quote again bits and pieces from the design stories workshop), like research, is an inherently communal endeavor, which I am overwhelming grateful to be a part of. I am similarly grateful to be presenting something so methodologically, topically, and personally meaningful as my first, first-author conference proceedings as a PhD student. Please feel free to reach out to me via email (faye.kollig@colorado.edu); I would love to hear to any critiques, questions, musings, entanglements, or stories of your own.