In-context learning (ICL) has proven to be adept at adapting large language models (LLMs) to downstream tasks without parameter updates, based on a few demonstration examples. Prior work has found that the ICL performance is susceptible to the selection of examples in prompt and made efforts to stabilize it. However, existing example selection studies ignore the ethical risks behind the examples selected, such as gender and race bias. In this work, we conduct extensive experiments and discover that (1) example selection with high accuracy does not mean low bias; (2) example selection for ICL may amplify the biases of LLMs; (3) example selection contributes to spurious correlations of LLMs. Based on the above observations, we propose the Remind with Bias-aware Embedding (ReBE), which removes the spurious correlations through contrastive learning and obtains bias-aware embedding for LLMs based on prompt tuning. Finally, we demonstrate that ReBE effectively mitigates biases of LLMs without significantly compromising accuracy and is highly compatible with existing example selection methods.

With the rapid deployment of Chinese large language models (LLMs), culturally-grounded bias evaluation remains understudied due to the dominance of English benchmarks and simplistic Chinese scenarios. To address this, we propose GeWu, a comprehensive benchmark featuring a culturally-aware dataset of 60,192 questions spanning 14 social groups with fine-grained Chinese contexts, significantly exceeding existing resources in breadth and depth. Our two-stage evaluation first quantifies bias via multiple-choice questions using a novel probability-based scoring mechanism to sensitively capture bias tendencies, distilling high-bias scenarios into GeWu-1K. This refined subset then enables multi-turn dialogue evaluations for in-depth analysis under realistic conditions. Experiments reveal that GeWu effectively exposes social biases in state-of-the-art Chinese LLMs, with 13.93% of scenarios eliciting universal bias across all models. This highlights persistent challenges and provides actionable insights for bias mitigation in Chinese contexts.

Multisensory stimulation promises in improving older adults’ affective experiences, yet its effectiveness depends on seamless affective congruency across sensory cues. This study investigated how visual, auditory, and kinetics correspondence and congruency shape affective experiences through two experiments. Experiment I examined timbre–color associations, showing that affective alignment strengthens perceived correspondence. Experiment II explored auditory–kinetics synchrony in a cross-modal art system, revealing no significant differences across conditions but indicating that older adults with lower cognitive abilities reported higher pleasure than higher-ability peers. Building on these results, an artificial intelligence (Al)-infused mode was integrated to transform strokes into real-time ink-style artworks, reducing cognitive effort, sustaining engagement. Findings demonstrate that AI enhances positive affect (pleasure, surprise, valence, and arousal) and mitigates negative affect (sadness, anger), with effects maximized by high sensory synchrony, providing compensatory support for users with lower cognitive abilities. These findings inform multisensory system design for older adults’ cognitive and affective needs.

In interpersonal interactions, individuals often exhibit avoidance behaviors toward others they find unpleasant, which can undermine the comfort of everyday social experiences. Existing human-computer interaction (HCI) research has primarily focused on promoting social connections, while support for avoidance-oriented social situations remains underexplored. To address this gap, we propose leveraging Diminished Reality (DR) technology to obscure perceptual cues of undesirable individuals. We designed and implemented a mixed reality prototype system and conducted experiments manipulating both the occlusion method and social distance. Results indicate that DR significantly reduces users’ social anxiety and sense of social presence. Moreover, participants generally expressed positive attitudes toward usage intention and ethical considerations. This work extends HCI research on social comfort, shifting the focus from “facilitating connection” to “supporting avoidance”.

Chinese Kung Fu not only emphasizes external techniques and movements but also places great importance on the cultivation of internal states such as Fa Li (force exertion) and Qi (energy control). However, existing motor learning systems predominantly focus on improving movement accuracy, with limited attention to the awareness and guidance of these internal states. Kung Fu films often use visual effects (VFX) to vividly express traditional cultural imagery. Inspired by this, we designed and developed a wearable motor learning system and interactive interface that integrates interactive VFX and haptic feedback to enhance users’ awareness of internal states such as force exertion and Qi, thereby improving both learning effectiveness and user experience. Through prototyping and an exploratory study, we found that the system significantly improved users’ awareness of Qi and force exertion, as well as their overall training experience. In addition, we identified several usability issues and proposed corresponding design improvements. This study introduces a novel visualization framework for internal state cues in Kung Fu training, offering new perspectives and practical approaches for the design of motor learning systems.

Yuxuan Li, Xinwei Guo, Jiashi Gao, Guanhua Chen, Xiangyu Zhao, Jiaxin Zhang, Quanying Liu, Haiyan Wu, Xin Yao, Xuetao Wei. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Social cues play a critical role in the exchange of information during patient–physician communication. Although previous research has recognized the impact of social cues on patients’ perceptions and behaviors in mobile medical consultations, little is known about how different combinations of social cues affect the interactions. Employing a within-subject factorial design experiment, this study examined the effects of three social cues—empathic cues, interactivity cues, and social validation—on patients’ perceptions of interactivity, social presence, and trust. A total of 103 participants was recruited. The results revealed the various significant impacts of empathic cues, interactivity cues, and social validation on perceived interactivity, and social validation affected trust, as well as the relationships among these perceptions. The findings suggested strategies for integrating various social cues to enhance patients’ positive perceptions, offering potential benefits to patient–physician communication in mobile medical platforms.

Affective robots can elicit psychological responses such as attachment and intimacy, which may help alleviate anxiety and enrich users’ emotional experiences. While such robots can take the form of agents (controlled by algorithms) or avatars (controlled by humans or animals), the differential effects of these embodiment forms on users’ emotional responses remain underexplored, particularly in scenarios where avatars are controlled by animals. In this study, we conducted a Wizard of Oz experiment to compare the emotional experience and anxiety relief provided by a robotic cat under different embodiment forms and interaction modes (unidirectional vs. bidirectional). The results indicate that the avatar embodiment significantly enhances users’ affective experiences, fostering stronger emotional bonds and more effective anxiety relief. However, no significant differences were found between the interaction modes with respect to either anxiety relief or emotional outcomes. These findings offer valuable insights for the design of emotionally engaging embodied intelligent systems in contexts such as emotional companionship and mental health interventions. • A wearable robot prototype enables remote interaction between humans and real cats. • Robot embodiment forms and interaction modes affect users’ anxiety and affective experience. • The avatar form leads to greater anxiety relief and stronger affective experiences. • No significant emotional differences are found between interaction modes.

While mobile payment services have been flourishing in China, users have continually questioned the security of these transactions. Although customization has been proposed as a vital factor for mobile commerce, minimal knowledge exists regarding how it affects users’ perceived security in mobile payment transactions. A quantitative diary study was therefore conducted to provide insight into the personality traits that motivate customization behaviors in security, and how such behaviors influence perceived security under different use contexts in relation to mobile payments. First, an instrument for the diary study was developed through an interview. Then, 134 responses from mobile payment users were used to examine the relationships between personality traits and customization behaviors. Among them, the diary was completed by 67 mobile payment users who reported their perceived security for 1094 recoded payment events across various use contexts for periods ranging between 5 and 15 days. The results showed that the personality traits of extraversion and intellect influence users’ customization behaviors and these behaviors have a positive effect on perceived security. Additionally, the relationship between customization behaviors and perceived security was moderated by the task and technical contexts. Based on these findings, design implications and opportunities for mobile payment services are described.