Learning a skill is a complex cognitive task that requires processing multiple sensorimotor information. To support the skill learners, in this study, we explored and compared how to exploit AI wisely to teach skills. We selected engraving as the target skill and implemented it in a haptic-audio-visual (HAV) VR environment. As the haptic device, we used a desktop force-feedback device to render the movement and force profile of the tooltip. Starting from gathering experts’ motion and force profile data, we designed a Long Short-Term Memory (LSTM)-based AI model that discriminates the user's behavior and status of the tooltip using the acquired data. With the VR environment and the AI model, we evaluated four teaching strategies for haptic dexterous skill transfer– Guidance, Disturbance, Hybrid , and Adaptive feedback . Guidance feedback provides force cues to help the users, whereas Disturbance applies force that prevents the user's movement. Hybrid alters its mode between Guidance and Disturbance based on the user's performance. Lastly, Adaptive feedback modulates the threshold for altering the mode of Hybrid based on the user's history. We conducted a user experiment with seven sessions: one pre-test, three training sessions, one immediate retention, and two delayed retention sessions. In the results, we found: 1) Guidance showed a steep learning curve during the training sessions, but the participants lost the learning effect in the retention sessions, 2) the learning with Hybrid was the slowest but remained longer, even showed better performance in delayed retention tests, and 3) Adaptive strategy exhibited superior long-term learning effects compared to Hybrid and further suggested improved performance stability. These results seem to follow both the guidance hypothesis and the phenomenon of offline consolidation in learning, highlighting the importance of designing adaptive strategies in Teaching AI systems for dexterous skill learning.
Human-AI collaboration is increasingly relevant in consequential areas where AI recommendations support human discretion. However, human-AI teams’ effectiveness, capability, and fairness highly depend on human perceptions of AI. Positive fairness perceptions have been shown to foster trust and acceptance of AI recommendations. Yet, work on confirmation bias highlights that humans selectively adhere to AI recommendations that align with their expectations and beliefs—despite those recommendations not being necessarily correct or fair. This raises the question of whether confirmation bias also transfers to the alignment of gender bias between human and AI decisions. In our study, we examine how gender bias alignment and the presence of explanations accompanying the AI recommendation influence fairness perceptions, human self-confidence, and reliance on the AI system. Findings from two 2 \(\times\) 2 between-subjects studies underscore the relationship between gender-bias alignment, fairness perceptions, and reliance, demonstrating that merely constructing a “formally fair” AI system is insufficient for optimal human–AI collaboration. Ultimately, AI recommendations are likely to be overridden when the system's biases do not align with those of its users.
As robot programming becomes increasingly important in enabling non-expert users to customize autonomous systems across domains such as manufacturing, healthcare, and service industries, there is a growing demand for methods that are both intuitive and support users’ perceived trust. While natural language programming (NLP) lowers the barrier to entry by allowing users to describe robot behavior in everyday language, it often lacks structural transparency—particularly when paired with opaque deep learning models. To address this challenge, we present LLiBTree, a novel robot programming framework that integrates Large Language Models (LLMs) directly into Behavior Trees (BTs) as executable nodes. This approach allows users to iteratively define and refine robot behavior via natural language while preserving the modularity and clarity of visual programming. By embedding the LLM within the BT, users can generate and replace sub-trees in response to natural language instructions, enabling a seamless blend of language-based flexibility and structured visual logic. We evaluate LLiBTree through a controlled user study, comparing it with both natural language-only and conventional visual programming methods. The results show that the PreGeneratedLLiBTree condition yields significantly higher perceived trust, while LLiBTree conditions achieve usability that is non-inferior to natural language-only programming and improve users’ sense of agency, suggesting that LLiBTree offers a practical approach for end-user robot programming.
Recent research has demonstrated the potential for representing intelligent guidance using multi-modal cues, yet few guidelines or processes exist to guide the design of such a system. In this article, we seek to address this gap by investigating the design of multi-modal assistant systems for setting the optimal parameters in industrial plants. We present the results of two user studies conducted with a total of 37 participants to evaluate the effectiveness and experience of different combinations of visual ( Highlights and Ambient lights) and haptic ( Clicks and Vibration ) modalities for providing intelligent dynamic guidance. Our findings demonstrate that providing the intelligent guidance with the multi-modality of Highlights+Ambient resulted in shorter task duration and higher practicality than Ambient lights alone. Regarding task accuracy, the Highlights and Vibration outside-zone guidance resulted in no failed attempts. Moreover, Highlights+Ambient+Vibration guidance was rated as having lower usability than Highlights+Ambient , and higher mental demand than merely Highlights .
Human-in-the-loop methods leverage human feedback to enhance machine learning and AI. Manual review of outputs can correct errors, identify model weaknesses, or expand labels to broaden model capabilities. Feedback collection methods range from simple flagging of outputs as correct or incorrect to more complex feature-level adjustments or natural language interpretations. This article presents a user study evaluating changes in user performance over time and explores the tradeoff between feedback quality and human effort. We compare four interactive input methods for reviewing and correcting outcomes in object detection and activity recognition in videos. Our findings indicate that while some complex input methods, such as free-text, require more time, the quality and impact of their feedback on model accuracy often surpass those of simpler methods that require less effort. However, more effort does not always lead to better-quality feedback, especially when aiming to improve the model. Our VLM experiments show that the most accurate models were trained using detailed natural language feedback or precise word-level corrections, while simple yes/no judgments also led to solid performance at a much lower annotation cost.
