Computers in Human Behavior最新文献

Prior to the rapid growth of Artificial Intelligence (“AI”) in the consumer market, smart products received great attention from marketers and consumers. Given the recent increase in attention to AI technologies, this research explores consumers' preferences and intentions when products are framed as “smart” versus “AI-powered.” While previous literature has explored AI products and smart products individually, little is known about consumers’ preferences between the two products simultaneously. Three empirical experiments demonstrate that consumers show preference for products labeled as “smart” over those labeled as “AI-powered.” This preference is mediated by enhanced learning anxiety related to AI. The findings provide insights for marketers applying message framing, suggesting that labeling products as “smart” may evoke more favorable consumer behavior compared to the “AI-powered” label. Moreover, this research significantly contributes to the existing literature on perceptions and intentions related to AI and smart products by concurrently exploring consumer preferences regarding both smart products and AI-powered products.

The imminent arrival of Artificial General Intelligence (AGI) compels a reevaluation of AI-human interactions, particularly through affective communication. This research synthesizes insights from evolutionary biology, comparative psychology, and AI development, advocating for a paradigm shift beyond conventional human-like cognitive processes. It emphasizes the universal nature of affective pathways, as evidenced across various species. We introduce three affective interaction models — the Affective Threshold Model, the Dynamic Set-Point Model, and the Affective Schema Model — all of which stem from an in-depth analysis of interspecies communications. These models present a roadmap to craft AI interfaces attuned to human affective experiences, elucidating avenues of trust, intuition, and reciprocal recognition between machines and their human counterparts. By further crystallizing the concept of the “Large Affect Model”, we project a horizon where AI not only deciphers but also empathizes with human partners, paving the way for a revolutionary cooperative paradigm between AI and humanity.

In spite of the common disadvantages, including user addiction and so on, the green online gaming system has a potential positive advantage to foster offline pro-environmental behavior through the special online experience. We developed a theoretical framework synthesizing motivation theory, flow theory, dual-system theory, and user engagement theory, aimed at exploring the mechanism between green gaming systems and actual pro-environmental behaviors. Based on the analysis of data from 364 participants utilizing Ant Forest across China, the results showed that motivations generated by using green online gaming systems indirectly promote pro-environmental behavior. The underlying psychological processes include the flow experience, social overload, and green engagement. Especially, user green engagement serves as a critical bridge linking virtual activities to real-world environmental behavior. Furthermore, perceived authenticity significantly facilitates the transition of online experiences into real-world pro-environmental behavior. This research not only advances the theoretical understanding of gamification and pro-environmental behavior but also provides practical references for promoting pro-environmental behavior effectively in managing green gaming systems.

Online video social interaction is now commonplace following rapid technological advances and the Covid-19 pandemic. Whether computer mediated communication (CMC) fundamentally changes nonverbal behaviour and social responses from others is unknown. Here, we conducted a repeated measures experimental study (N = 66) comparing three types of dyadic social interactions: in person, online video call (with self-view) and online video call (no self-view). Facial videos were analysed using automated facial movement tracking (based on the Facial Action Coding System: FACS). Independent raters made first impression judgements across all conditions (N = 198). Overall, people were more facially expressive in person compared to CMC, but there were significant individual differences across participants. Agreeableness was associated with a particular increase in expressivity in person compared to online, while extroversion was associated with greater expressivity in online video calls, but only when self-view was visible. Older adults were most impacted by CMC and showed the greatest reduction in facial expressivity online compared to in person. The first impressions of observers did not differ as a function of CMC. These results suggest that CMC does alter facial expressivity during social interaction, but that there is an important interplay with individual differences.

Through creating an environment rich in computational and communication capabilities, ubiquitous computing gradually integrates it with human activities. Inspired by adaptive ubiquitous learning, various intelligent devices (e.g., roadside units and infrared sensors) deployed in the Internet of Vehicles (IoV) are expected to be critical to mitigating urban traffic congestion and enhancing travel safety. In addition, benefiting from the advantages of high mobility and real-time response, Unmanned Aerial Vehicles (UAVs) embody substantial prospects to assist IoV in efficiently and flexibly handling latency-sensitive, computation-intensive tasks. Nevertheless, due to time-varying demands and heterogeneous computing resources, it is challenging to provide effective service for mobile devices while guaranteeing high-quality data transmission. Therefore, a distributed service offloading system framework in UAV-enhanced IoV is designed. To minimize the service latency, a game theory-based distributed service offloading algorithm, named G-DSO, is proposed to realize adaptive ubiquitous learning for service request distribution. Finally, numerous experiments are implemented based on real-world service requirement datasets. Experimental results demonstrate that the proposed G-DSO approach improves the hit rate by 2.68% to 74.42% compared with four existing service offloading methods, verifying the effectiveness and good scalability of G-DSO.

The Internet of Vehicles (IoV) has the potential to bring about a revolutionary transformation in transportation through its influence on human behavior and interactions between users and vehicles. However, interoperability challenges between retailer organizations and manufacturers present a barrier to decision-making processes and impact the human-centric nature of the IoV. Ethical dilemmas arise as a result of the IoV’s inability to prevent accidents, particularly in critical situations. This study aims to enhance the IoV’s effectiveness by carefully selecting and improving essential attributes from various data sources, including sensors, GPS, 5G or 6G communication networks, and real-time data provisioning. To achieve the aim of the proposed study, a Multi-criterion Decision-making (MCDM) approach is proposed, which allows for the analysis and selection of optimal choices while taking into account various quantitative and qualitative factors. Despite the challenges posed by complex models and ambiguous data, MCDM remains an indispensable technique for aligning transportation systems with current expectations. The CRITIC and TOPSIS MCDM-enabled methodologies are employed to analyze IoV architecture, prioritizing significant elements that impact system performance and identifying optimal solutions by considering complications from worst-case scenarios. The study will assist engineers, scientists, and organizations to develop smart IoV systems that will cater to human needs by improving mobility and inspiration among users.

Automated vehicles (AVs) provide numerous advantages over manually operated vehicles, but the extent of these benefits depends on whether we engage with AVs safely and efficiently. To achieve such interactions with AVs, an appropriate – or calibrated – level of trust in AVs especially during critical scenarios, is fundamental. The trust level also impacts individuals' decisions regarding the utilisation of AV technology. This study investigates trust calibration and factors that influence how trust develops in AVs compared to human drivers. Two groups of participants underwent a driving simulation, experiencing either a ride in a human-driven taxi or an AV, during which a dangerous situation occurred. Before, during and after the simulation, the passengers’ trust was measured.

Pre-simulation trust was higher in the human driver than in the AV, but this difference disappeared after the simulation. Noticeably, during the simulation trust did not differ between the groups. Instead, the critical situation significantly influenced trust: following the dangerous incident, trust levels in both conditions dropped but recovered until the simulation ended. Additionally, self-esteem, which has been associated with trust in the past, was investigated. However, no significant relationship between self-esteem and trust was found in this study. Overall, the findings indicate that the dangerous situation prompted heightened caution among participants. A process of trust calibration was initiated in which the participants’ trust was highly susceptible to the driving style of the driver/AV. Moreover, the comparable evolution of trust in the human-driven vehicle and the AV, sheds light on the dynamics underlying attitudes towards AVs.

This study used video-based instruction to promote adolescents' online credibility evaluation skills and to examine whether storification of instructional videos can maintain students' situational interest and positive activating epistemic emotions during a four-lesson intervention better than non-storified instructional videos. The study also investigated whether storified instruction can enhance students' learning of online credibility evaluation skills better than non-storified instruction. The learning content of the videos was equivalent, differing only in terms of storification. Students participated in either a non-storified (n = 135) or storified (n = 115) video-based instruction (4 x 45-min lesson). In each of the first three lessons, students watched an instructional video explaining one credibility evaluation strategy. In the fourth lesson, the video introduced a case requiring students to apply the taught strategies. After watching the video, students' situational interest and positive activating epistemic emotions (i.e., curiosity and excitement) were measured. Then, students practiced the taught strategies with learning tasks. Students' credibility evaluation skills were measured before and after the intervention with an online credibility evaluation task. The latent growth curve model showed that non-storified and storified videos similarly maintained students' situational interest. No decline was observed. However, a small decline in positive activating epistemic emotions was observed in both instructional conditions. Students' credibility evaluation skills improved equally across instructions. The results underline that professionally produced videos following several multimedia design principles can trigger and maintain situational interest and enhance students’ online credibility evaluation skills without the need to incorporate additional storified elements.

This study explores the cognitive load and learning outcomes associated with using large language models (LLMs) versus traditional search engines for information gathering during learning. A total of 91 university students were randomly assigned to either use ChatGPT3.5 or Google to research the socio-scientific issue of nanoparticles in sunscreen to derive valid recommendations and justifications. The study aimed to investigate potential differences in cognitive load, as well as the quality and homogeneity of the students' recommendations and justifications. Results indicated that students using LLMs experienced significantly lower cognitive load. However, despite this reduction, these students demonstrated lower-quality reasoning and argumentation in their final recommendations compared to those who used traditional search engines. Further, the homogeneity of the recommendations and justifications did not differ significantly between the two groups, suggesting that LLMs did not restrict the diversity of students’ perspectives. These findings highlight the nuanced implications of digital tools on learning, suggesting that while LLMs can decrease the cognitive burden associated with information gathering during a learning task, they may not promote deeper engagement with content necessary for high-quality learning per se.