Jessi Stark

HCI Researcher

PhD Candidate at the University of Toronto

This website is undergoing a move so some information might be missing. In the meantime, you can go here to find it.

Greetings! Let me tell you a bit about what I do.

As a human-computer interaction researcher, I think a lot about how humans use technology. These days, I am particularly interested in how creative writers engage with writing support tools and how such tools can be designed to better support creative writers.

In the past, my research focused on creativity support tools for novice makers in makerspaces. I explored multi-perspective video interfaces to support 3D assembly tasks, and instrumented a makerspace to increase the situation awareness of novices in the space.

Since Fall 2020, I have been a PhD student in the Computer Science department at the University of Toronto, supervised by Dr. Daniel Wigdor. In 2020, I also completed my MSc in Computer Science at the University of Calgary, supervised by Dr. Ehud Sharlin and Dr. Anthony Tang. I received my BSc in Computer Science from the University of Calgary in 2017, during which I dabbled in human-robot interaction research. Before entering the world of computer science, I studied journalism and embarked on a brief and morbid career editing obituaries.

My Research

My research delves into the dynamics between creative fiction writers and AI-driven writing support tools, focusing on mitigating frictions that arise while ensuring the writer's sense of ownership over the text remains intact. Specifically, I explore the potential of Large Language Model-based writing tools (LLMs) to assist fiction writers without encroaching on their creative autonomy. In my upcoming study, I will investigate innovative approaches leveraging LLMs to enhance fiction writing experiences through the development of a Google Docs add-on, currently under construction. This tool boasts features such as character profiling and relationship mapping, enabling writers to maintain consistency and coherence within their narratives. Derived from insights garnered in previous research, these features address identified writing challenges and preserve the writer's sense of ownership. My implementation of these functionalities involves adept manipulation of prompts to optimize LLM outputs, task decomposition to manage model responses, and robust handling of unpredictable model outputs.

Publications

Full papers

Workshop submissions

Posters

CV

Read it here.

Contact

Email me at jtstark@dgp.toronto.edu