While the potential of Artificial Intelligence (AI)—particularly Natural Language Processing (NLP) models—for detecting symptoms of depression from text has been vastly researched, only a few studies examine such potential for the detection of social anxiety symptoms. We investigated the ability of the large language model (LLM) GPT-4 to correctly infer social anxiety symptom strength from transcripts obtained from semi-structured interviews. N = 51 adult participants were recruited from a convenience sample of the German population. Participants filled in a self-report questionnaire on social anxiety symptoms (SPIN) prior to being interviewed on a secure online teleconference platform. Transcripts from these interviews were then evaluated by GPT-4. GPT-4 predictions were highly correlated (r = 0.79) with scores obtained on the social anxiety self-report measure. Following the cut-off conventions for this population, an F₁ accuracy score of 0.84 could be obtained. Future research should examine whether these findings hold true in larger and more diverse datasets.

Social anxiety is both highly prevalent and debilitating, with dysfunctional beliefs playing a crucial role in its persistence. This randomized controlled trial examined whether a remote avatar-based intervention for restructuring dysfunctional beliefs was associated with changes in social anxiety symptoms in a sample with subclinical social anxiety disorder (SAD). Data from N = 235 participants who completed all study stages were included in the analyses. While participants in the intervention group were tasked with contradicting their personal dysfunctional beliefs verbalized by the avatar, participants in the control group contradicted nonsense statements uttered by the avatar. Three sessions took place over three consecutive days. Data collection occurred pre-, post-intervention, and at a 14-days-follow-up. A significant group x time interaction effect ( F (1, 233) = 4.106, p < .05) indicates a stronger SAD symptom reduction within the intervention group in comparison to the control group. Post-hoc analysis revealed a medium effect size for symptom reduction in the intervention group ( d = 0.57) and a small effect size for the control group ( d = 0.34). An additional linear multiple regression analysis identified that increased self-esteem ( B = − 0.56, p < .001) was associated with reductions in SAD symptoms. An increase in dysfunctional social phobia cognitions was associated with an increase in SAD symptoms ( B = 0.175, p = .021). Avatar-based cognitive restructuring was associated with reductions in SAD symptoms in a subclinical sample. Changes in self-esteem and social phobia-related cognitions were also associated with symptom changes, but causal relationships cannot be inferred from this study design.

AI-based recruiting is emerging as a critical tool for companies to attract and engage candidates. In this interdisciplinary study, we present a framework for AI-based active sourcing in recruitment to explore opportunities for incorporating considerations of contestability, i.e., the openness of an AI system to human intervention by those affected. The proposed framework is structured around four key modules: Active searching, skills extraction, skills matching, and automated and personalized approach. After introducing the design and functionality of each module, we critically examine the associated opportunities and challenges regarding contestability, including their connection to other ethical aspects like transparency. We conclude with a discussion of pertinent challenges and points of concern, as well as potential practical solutions to enhance contestability and mitigate ethical risks. Our work aims to explore contestability in the context of responsible, ethically acceptable development and application of AI-driven active sourcing systems in human resource management. Future research should empirically assess the integration of contestability aspects in active sourcing approaches in practice.

Aim: The primary objective of this study was to examine and explain the public panic consumption model based on the stimulus–organism–response theory, during the peak time period of the COVID-19 pandemic in China. Subject and methods: The research data were collected through the questionnaires adapted for the purpose of this survey, which included a total of 408 participants (33% female) from the global population. The stepwise regression analysis has been conducted. Results: The results have shown that both physical social networks and online social networks have a significant positive impact on infection risk perception, but the impact of physical social networks has proved to be a better predictor. Infection information obtained from physical social networks affects conformity buying and uncontrolled self-medication if the perceived risk perception is higher. When it comes to the impact of risk perception on conformity buying, health change has a significant negative regulatory effect. In the impact of risk perception on the uncontrolled self-medication, health change has no significant regulatory effect. Conclusions: During the recent epidemic states, it is noticeable that public panic consumption has aggravated the difficulty of emergency management. Especially when it comes to the management of medical materials and medicines. In order to tackle this challenge in the future, the trigger mechanisms behind panic consumption are revealed.