Intelligent Reasoning Cues: A Framework and Case Study of the Roles of AI Information in Complex Decisions

TL;DR

This work rethinks AI-enabled decision support by framing interfaces as collections of intelligent reasoning cues that can shape clinicians' reasoning rather than merely offering a single recommendation. Through two ICU sepsis studies (contextual inquiry with six care teams and think-aloud sessions with 25 physicians) and a transformer-based nearest-neighbor embedding of patient states, the authors define eight cues across case features, risk predictions, peer actions, and plans, implemented in seven interfaces. They identify 11 distinct patterns by which cues influence reasoning—ranging from Considering Plans to Plan Preference—and show that perceived usefulness hinges on tasks’ variability, compatibility with information needs, and complementarity with guidelines. The findings yield design recommendations for sepsis and other high-stakes domains, including focusing on discretionary tasks, enabling adaptive cues, and leveraging true-data evidence to support nuanced decisions, with implications for GenAI-enabled decision support. Overall, the intelligent reasoning cues framework offers a generalizable, mechanism-grounded approach to crafting AI interfaces that effectively scaffold expert reasoning in complex decision settings.

Abstract

Artificial intelligence (AI)-based decision support systems can be highly accurate yet still fail to support users or improve decisions. Existing theories of AI-assisted decision-making focus on calibrating reliance on AI advice, leaving it unclear how different system designs might influence the reasoning processes underneath. We address this gap by reconsidering AI interfaces as collections of intelligent reasoning cues: discrete pieces of AI information that can individually influence decision-making. We then explore the roles of eight types of reasoning cues in a high-stakes clinical decision (treating patients with sepsis in intensive care). Through contextual inquiries with six teams and a think-aloud study with 25 physicians, we find that reasoning cues have distinct patterns of influence that can directly inform design. Our results also suggest that reasoning cues should prioritize tasks with high variability and discretion, adapt to ensure compatibility with evolving decision needs, and provide complementary, rigorous insights on complex cases.

Figures (4)

• Figure 1: Our framework models AI interfaces as collections of multiple intelligent reasoning cues, shown here on the Interactive Treatment Risk interface from our study. Each reasoning cue (labeled by codes R1-R8 and described in Sec. \ref{['sec:reasoning-cue-desc']}) conveys to the user a discrete insight that could support a decision.
• Figure 2: (a) Intelligent reasoning cues, numbered R1-R8, considered in this study. (b) AI interfaces, each containing multiple reasoning cues. Treatment Risk and Mortality Risk refer to different variants of the same interfaces with different predictive targets.
• Figure 4: Example of the interface and vignettes shown to participants in Study 2. The blue box labeled Sepsis AI would contain the AI interface and reasoning cues they were randomized to receive for the given patient.
• Figure 5: A theme generated from the first round of affinity mapping of codes for each patient case. This example is from Richard's patient case.