Mobile GUI agents often act confidently under ambiguous or incomplete user instructions-a common situation in real-world use. Even when explicitly offered a "call the user" action, open-weight agents ask for clarification in only 0.1% of cases on average, instead guessing the user’s intent and proceeding with actions. We name this failure Act-without-Asking. Existing step-level benchmarks largely miss this failure mode because they credit any ground-truth-matching action regardless of asking. We introduce MID-Bench, a step-level ambiguity benchmark constructed by rewriting clear instructions into ambiguous ones, with a strict clarification protocol: a user simulator answering only what is asked, and a scoring rule crediting a step only when the agent both asks an appropriate question and emits the correct final action. The Hard subset applies stricter masking to the action-critical detail in the instruction, making it harder to succeed without asking. Across four open-weight GUI agents, train-free cascades in which a gate decides whether to act or ask the user for clarification substantially improve ambiguous-task performance while preserving accuracy on clear instructions; LoRA fine-tuning on MID-Bench further combines effectively with these cascades.

The increasing demand for mental health services has led to the rise of AI-driven mental health chatbots, though challenges related to privacy, data collection, and expertise persist. Motivational Interviewing (MI) is gaining attention as a theoretical basis for boosting expertise in the development of these chatbots. However, existing datasets are showing limitations for training chatbots, leading to a substantial demand for publicly available resources in the field of MI and psychotherapy. These challenges are even more pronounced in non-English languages, where they receive less attention. In this paper, we propose a novel framework that simulates MI sessions enriched with the expertise of professional therapists. We train an MI forecaster model that mimics the behavioral choices of professional therapists and employ Large Language Models (LLMs) to generate utterances through prompt engineering. Then, we present KMI, the first synthetic dataset theoretically grounded in MI, containing 1,000 high-quality Korean Motivational Interviewing dialogues. Through an extensive expert evaluation of the generated dataset and the dialogue model trained on it, we demonstrate the quality, expertise, and practicality of KMI. We also introduce novel metrics derived from MI theory in order to evaluate dialogues from the perspective of MI.

Recent advances in large language models and vision-language models have led to growing interest in explainable evaluation metrics for image captioning. However, these metrics generate explanations without standardized criteria, and the overall quality of the generated explanations remains unverified. In this paper, we propose EXPERT, a reference-free evaluation metric that provides structured explanations based on three fundamental criteria: fluency, relevance, and descriptiveness. By constructing large-scale datasets of high-quality structured explanations, we develop a two-stage evaluation template to effectively supervise a vision-language model for both scoring and explanation generation. EXPERT achieves state-of-the-art results on benchmark datasets while providing significantly higher-quality explanations than existing metrics, as validated through comprehensive human evaluation. Our code and datasets are available at https://github.com/hjkim811/EXPERT.