Some selected publications are below. For a more complete list, see my Google Scholar profile. For less-polished but more recent thoughts, see the blog posts here.
Large language model (LLM)-based writing tools, such as ChatGPT, Bing Chat, and Bard, promise efficiency and productivity; however, they can overshadow the thinking process necessary for original writing, compromising autonomy and authorship. This limitation is not intrinsic to LLM itself but is due to the design of the chatbot user interface, which encourages reliance on a LLM for both thinking and writing. In order to explore alternative approaches and the potential benefits that LLMs can offer to the writing process, we have implemented a user interface that encourages writers to reflect on texts they wrote themselves. By conducting a pilot study, we investigated the impact of new user interface affordances on encouraging writers to reflect on their writing. Through the analysis of feedback from study participants, we identified ways to improve the interface and interaction, as well as an opportunity to incorporate conversational aspects into our prototype. Additionally, we outlined a potential future research plan to conduct a deployment study to observe user interactions with our interface in a natural setting.
No matching items