Early language development shapes children’s later literacy and learning, yet many families have limited access to scalable, high-quality support at home. Recent advances in generative AI make it possible for social robots to move beyond scripted interactions and engage children in adaptive, conversational activities, but it remains unclear how to design such systems for pre-schoolers and how children engage with them over time in the home. We present ELLA (Early Language Learning Agent), an autonomous, LLM-powered social robot that supports early language development through interactive storytelling, parent-selected language targets, and scaffolded dialogue. Using a multi-phased, human-centered process, we interviewed parents (n=7) and educators (n=5) and iteratively refined ELLA through twelve in-home design workshops. We then deployed ELLA with ten children for eight days. We report design insights from in-home workshops, characterize children’s engagement and behaviors during deployment, and distill design implications for generative AI–powered social robots supporting early language learning at home.

Large language models (LLMs) enabled conversational robots to shift toward free-form interaction. However, without context-specific adaptation, generic LLM outputs can be ineffective or inappropriate. This adaptation is often attempted through prompt engineering, which is non-intuitive and tedious. Moreover, predominant design practice in HRI relies on impression-based, trial-and-error refinement without structured methods or tools, making the process inefficient and inconsistent. To address this, we present AI-Aided Conversation Engine (ACE) to support deliberate design of human-robot conversations with three key innovations: 1) an LLM-powered voice agent that scaffolds initial prompt creation to overcome the "blank page problem," 2) an annotation interface that enables the collection of granular and grounded feedback on conversational transcripts, and 3) using LLMs to translate user feedback into prompt refinements. We evaluated ACE through two user studies, examining both designs' experience and end users' interactions with robots designed using ACE. Results show that ACE facilitates the creation of robot behavior prompts with greater clarity and specificity, and that the prompts generated with ACE lead to higher-quality human-robot conversational interactions.

The integration of large language models (LLMs) into conversational robots has made human-robot conversations more dynamic. Yet, LLM-powered conversational robots remain prone to errors, e.g., misunderstanding user intent, prematurely interrupting users, or failing to respond altogether. Detecting and addressing these failures is critical for preventing conversational breakdowns, avoiding task disruptions, and sustaining user trust. To tackle this problem, the ERR@HRI 2.0 Challenge provides a multimodal dataset of LLM-powered conversational robot failures during human-robot conversations and encourages researchers to benchmark machine learning models designed to detect robot failures. The dataset includes 16 hours of dyadic human-robot interactions, incorporating facial, speech, and head movement features. Each interaction is annotated with the presence or absence of robot errors from the system perspective, and perceived user intention to correct for a mismatch between robot behavior and user expectation. Participants are invited to form teams and develop machine learning models that detect these failures using multimodal data. Submissions will be evaluated using various performance metrics, including detection accuracy and false positive rate. This challenge represents another key step toward improving failure detection in human-robot interaction through social signal analysis.

Appropriate reliance is critical to achieving synergistic human-AI collaboration. For instance, when users over-rely on AI assistance, their human-AI team performance is bounded by the model's capability. This work studies how the presentation of model uncertainty may steer users' decision-making toward fostering appropriate reliance. Our results demonstrate that showing the calibrated model uncertainty alone is inadequate. Rather, calibrating model uncertainty and presenting it in a frequency format allow users to adjust their reliance accordingly and help reduce the effect of confirmation bias on their decisions. Furthermore, the critical nature of our skin cancer screening task skews participants' judgment, causing their reliance to vary depending on their initial decision. Additionally, step-wise multiple regression analyses revealed how user demographics such as age and familiarity with probability and statistics influence human-AI collaborative decision-making. We discuss the potential for model uncertainty presentation, initial user decision, and user demographics to be incorporated in designing personalized AI aids for appropriate reliance.

Robotic systems designed to work alongside people are susceptible to technical and unexpected errors. Prior work has investigated a variety of strategies aimed at repairing people's trust in the robot after its erroneous operations. In this work, we explore the effect of post-error trust repair strategies (promise and explanation) on people's trust in the robot under varying power dynamics (supervisor and subordinate robot). Our results show that, regardless of the power dynamics, promise is more effective at repairing user trust than explanation. Moreover, people found a supervisor robot with verbal trust repair to be more trustworthy than a subordinate robot with verbal trust repair. Our results further reveal that people are prone to complying with the supervisor robot even if it is wrong. We discuss the ethical concerns in the use of supervisor robot and potential interventions to prevent improper compliance in users for more productive human-robot collaboration.

Human cognitive and decision-making abilities depreciate under pressure, motivating the emergence of artificial intelligence (AI) systems as decision support tools to assist people in performing tasks under stress. In this work, we study human decision-making behavior and task performance under time pressure---induced from limitedinitial observation time (time to perform the task before providing an initial response without AI input) andfinal decision time (time to weigh an AI's suggestion before reaching a collective human-AI team answer)---for spatial reasoning and count estimation tasks. Our results show that, while the impact of initial observation time on AI-assisted decision-making was dependent on task nature, participants were more likely to follow AI suggestions when they were provided with longer final decision time; moreover, although participants generally tended to adhere to their initial responses, they had more agency when they were more logically engaged in a task. Our results offer a nuanced understanding of human-AI collaboration under time pressure in different phases of the decision-making process.

Speech interfaces, such as personal assistants and screen readers, read image captions to users. Typically, however, only one caption is available per image, which may not be adequate for all situations (e.g., browsing large quantities of images). Long captions provide a deeper understanding of an image but require more time to listen to, whereas shorter captions may not allow for such thorough comprehension yet have the advantage of being faster to consume. We explore how to effectively collect both thumbnail captions—succinct image descriptions meant to be consumed quickly—and comprehensive captions—which allow individuals to understand visual content in greater detail. We consider text-based instructions and time-constrained methods to collect descriptions at these two levels of detail and find that a time-constrained method is the most effective for collecting thumbnail captions while preserving caption accuracy. Additionally, we verify that caption authors using this time-constrained method are still able to focus on the most important regions of an image by tracking their eye gaze. We evaluate our collected captions along human-rated axes—correctness, fluency, amount of detail, and mentions of important concepts—and discuss the potential for model-based metrics to perform large-scale automatic evaluations in the future.

Proper calibration of human reliance on AI is fundamental to achieving complementary performance in AI-assisted human decision-making. Most previous works focused on assessing user reliance, and more broadly trust, retrospectively, through user perceptions and task-based measures. In this work, we explore the relationship between eye gaze and reliance under varying task difficulties and AI performance levels in a spatial reasoning task. Our results show a strong positive correlation between percent gaze duration on the AI suggestion and user AI task agreement, as well as user perceived reliance. Moreover, user agency is preserved particularly when the task is easy and when AI performance is low or inconsistent. Our results also reveal nuanced differences between reliance and trust. We discuss the potential of using eye gaze to gauge human reliance on AI in real-time, enabling adaptive AI assistance for optimal human-AI team performance.