Large language models increasingly serve as judges in AI evaluation, but current approaches rely on consensus preferences that ignore individual evaluator variation. We propose a novel simulation approach that combines categorical judgments with evaluator-specific auxiliary data—retrospective reasoning traces and interface telemetry—to enable LLM-based simulation of individual evaluators via in-context learning. We conduct a systematic empirical study of this approach using multi-facet data from 32 trained annotators across 4,200 preference judgments in a 4 × 4 × 4 factorial design. Our key findings: (1) The simulation approach achieves up to 9.9 percentage point improvements over the Base Judge; (2) Reasoning traces provide the largest gains with higher collection efforts, while interface telemetry often hurts rather than helps performance despite being cheaper to collect. (3) Simulation difficulty is systematic, predicted by an evaluator’s neutral usage (most clearly on Helpfulness) and divergence from consensus; the neutral-usage tendency—rather than simulatability itself—is the cross-task-stable property (r = 0.728). These results establish both the potential and limits of evaluator-specific auxiliary data for personalized evaluation, offering methodological insights for scaling individual aware AI assessment.

Proactive AI writing assistants need to predict when users want drafting help, yet evidence about which contextual cues drive these preferences remains limited. Through a factorial vignette study with 50 participants making 750 pairwise comparisons in professional-email scenarios, we find that compositional effort is the strongest measured predictor (𝜌 = 0.597), whereas urgency has a near-zero scenario-level correlation and no reliable effect across inferential analyses. More critically, we observe a striking stated–revealed gap between abstract factor rankings and situated choices: they rank urgency first in self-reports despite it being the weakest behavioral driver, a contrast at the extremes even though mean stated urgency and effort ranks are close, i.e., 2.06 vs. 2.18. This misalignment has measurable consequences. Within this controlled professional-email testbed, the behavior-derived rule agrees with 61.3% of the 750 observed pairwise choices, compared with 57.7% for the stated-preference rule. This 3.6-point difference corresponds to a net gain of 27 matches. It also identifies a concrete design risk for proactive email assistance: stated rankings can place weight on cues that do not predict situated choices. Future studies should test whether stated rankings create similar design risks in other writing tasks.

Large Language Models (LLMs) are increasingly used for generating code solutions, empowered by features like self-debugging and self-reflection. However, LLMs often struggle with complex programming problems without human guidance. This paper investigates the strategies employed by expert programmers to steer code-generating LLMs toward successful outcomes. Through a study involving experts using natural language to guide GPT-4, Gemini Ultra, and, Claude 3.5 Sonnet on highly difficult programming challenges, we frame our analysis using the “Socratic Feedback” paradigm for understanding effective steering strategies. By analyzing 30 conversational transcripts across all three models, we map observed feedback strategies to five stages of Socratic Questioning: Definition, Elenhus, Maieutic, Dialectic, and Counter-factual reasoning. We find evidence that by employing a combination of different Socratic feedback strategies across multiple turns, programmers successfully guided the models to solve 74% of the problems that the models initially failed to solve on their own.