Qing (Nancy) Xia

“When should Generative AI (GenAI) assistants intervene, if at all?” While current GenAI assistants in knowledge work primarily operate reactively, responding only when users initiate an interaction, there is significant potential for these assistants to proactively assist in tasks. To tap into the potential of proactive GenAI assistants, it is important to examine when interventions are useful, when they are harmful, and how such judgements can be made reliably. The goal of this workshop, TimelyAI, at CHIWORK 2026 is to bring together Human-Computer Interaction (HCI) researchers interested in AI, psychological science, and social science to examine the timing of GenAI intervention in knowledge work and establish a shared research agenda.

Background: Instructional documents are crucial for safely operating medical devices. However, few studies have explicitly considered designing instructions for health care professionals (HCPs). Psychological research suggests that advanced learners with prior expertise in a domain are better able to generalize their existing knowledge to new areas compared to those with little prior expertise, resulting in different informational needs. Objective: We aimed to understand how HCPs learn from and use their existing expertise when interacting with instructions for use to familiarize themselves with a novel medical device. This would allow us to explore how instructional documents could be designed to better accommodate HCPs’ needs as experienced learners with prior knowledge in the domain. Methods: We conducted our studies centered around a novel liver support system, the HepatiCan and its current instructions for use. We conducted 3 user studies, first using semistructured interviews and think-aloud protocols to understand HCPs’ expectations for the process of setting up the HepatiCan and how training for the use of medical devices typically takes place. We followed this up with an observational study to understand HCPs’ actual experience in practice and corresponding instructional design needs. Results: Our results showed that HCPs’ prior expertise allowed them to prioritize key areas for attention, but also led them to make assumptions and potentially skim and miss important information. Visual representations and communication were preferred, as well as designs that supported memorization, as this was an essential function of using instruction manuals. Conclusions: Developers must be aware of what is considered “common” and potentially ignored knowledge within a specific domain in order to clearly emphasize important, safety-relevant information. We also recommend developers focus on using images with attached appendices where possible. This allows advanced users to process information more efficiently without compromising the needs of less familiar users for greater instructional details.

Spreadsheet collaboration provides valuable opportunities for learning and expertise sharing between colleagues. Sharing expertise is essential for the retention of important technical skillsets within organisations, but previous studies suggest that spreadsheet experts often fail to disseminate their knowledge to others. We suggest that social norms and beliefs surrounding the value of spreadsheet use significantly influence user engagement in sharing behaviours. To explore this, we conducted 31 semi-structured interviews with professional spreadsheet users from two separate samples. We found that spreadsheet providers face challenges in adapting highly personalised strategies to often subjective standards and in evaluating the appropriate social timing of sharing. In addition, conflicted self-evaluations of one's spreadsheet expertise, dismissive normative beliefs about the value of this knowledge, and concerns about the potential disruptions associated with collaboration can further deter sharing. We suggest that these observations reflect the challenges of long-term learning in feature-rich software designed primarily with initial learnability in mind. We therefore provide implications for design to navigate this tension. Overall, our findings demonstrate how the complex interaction between technology design and social dynamics can shape collaborative learning behaviours in the context of feature-rich software.