大学生AIGC健康信息甄别行为影响因素

随着以ChatGPT、DeepSeek为代表的生成式人工智能(AI)在全球范围内大规模流行，其在教育领域的应用已成为不容忽视的社会现象。本研究通过梳理大学生AI使用的现状，系统分析AI对大学生发展的双重影响：一方面，AI通过提供心理辅助服务、提升学习效率、赋能个性化发展等路径，为大学生心理建设注入正向动能；另一方面，其使用也存在心理依赖、学术创新弱化、隐私安全风险及价值观渗透等多元潜在风险挑战。研究最终提出“以人为本”的AI教育生态构建路径，包括课程体系完善、价值观引领、人机协同优化等维度，为规范大学生AI使用、守护心理健康提供理论与实践参考。

高校图书馆的数字阅读推广与信息素养教育存在密切交织的关系，都旨在提升用户的信息获取、处理、分析和应用能力，以适应数字化和信息化的社会需求。人工智能成为推动图书馆服务创新和转型升级的重要力量，数字阅读推广的智能化转型与信息素养教育的外延式拓展碰撞融合，基于融合发展理念，剖析探索数字阅读推广和信息素养教育融合发展的可行性路径，为新形势下高校图书馆信息服务职能和教育职能拓展提供新视角和新思路。

数字时代，人工智能技术(AI)既给学术研究与教学带来了极大的便利性也诱发了一定的风险，在高等教育中培养人工智能素养愈发关键。本研究基于扎根理论，通过质性分析软件Nvivo.15编码分析微博文本内容，旨在为高校图书馆优化人工智能素养教育路径提供指导和参考。研究发现，高校图书馆可从以下方面开展人工智能素养培养：明确AI时代核心能力、构建多元化教学实践框架、提升图书馆员服务能力以及加强学术伦理与风险教育，从而支持AI生态下的跨学科人才培养。

数字素养教育随着数字技术的发展不断更新迭代，生成式AI技术为数字素养教育提出新的机遇与挑战。文章首先阐述大学生数字素养内涵，接着分析生成式AI带来的机遇及面临的挑战。通过对广东省大学生的数字素养现状问卷调查与分析，提出优化策略以期提升大学生数字素养和适应时代发展需求。

随着互联网信息技术的不断发展，网络新媒体、大数据资源、云计算处理等信息技术成为了重要的教育工具，这为高校心理健康教育带来了新的契机。但互联网中充斥着大量良莠不齐的信息，处于人生重要成长阶段的高校大学生价值观发展并不成熟，更容易受到网络不良信息的影响造成认知偏差。因此互联网时代的到来不仅为高校心理健康教育带来了契机也提出了新的挑战和要求。本文主要通过从分析互联网时代落实心理健康教育创新的必要性以及在互联网时代背景下，心理健康教育面临的机遇和挑战入手，提出有效的运用互联网技术，创新高校心理健康教育的方法，让各种心理教育资源得到充分的利用，从而促使高校心理健康教育在新时代发展中最大程度地发挥育人成效。

人工智能已深度介入大学生知识习得与生产的全过程，对其网络素养的培育提出了更高的要求。当前，大学生网络素养存在显著的差异化特征：对AI信息的甄别能力不足、算法认知与批判意识薄弱、数据伦理与隐私保护观念欠缺、虚拟交往行为失范与数字公民意识模糊。与此同时，人工智能的快速发展也给大学生网络素养培育带来了新的契机：其出场逻辑重构了教育资源供给模式，交互逻辑革新了教学范式与认知路径，效能逻辑能实现精准培育与风险预警。基于此，文章提出从课程教学、实践训练与协同育人三个维度，构建“分层递进 + 螺旋上升”的课程教学机制、“人机协同 + 场景赋能”的实践训练机制，以及“家校社网”四位一体的协同联动机制，以系统化路径破解人工智能视域下大学生网络素养的培育困境。

随着科学技术的发展，人工智能已经在各大领域应用。其在科普出版领域的应用也越发广泛，无论是内容创作、编辑校对、媒体融合等都离不开人工智能的帮助。本文分析了人工智能在科普出版的应用现状以及在两者融合过程中所遇到的科学知识错误、知识产权保护、技术自身局限、编辑角色转变的挑战。针对这些问题，分别从技术监测、法律法规、技术成本、人员培养四个方面总结应对途径。提出了加强技术检测力度、建立完整法律标准、技术革新降低门槛、培养复合人才等措施。为科普出版的未来发展提供新路径，为科普出版的转型升级提供新措施。

随着ChatGPT技术在全球范围内迅速走红，新闻行业对其予以了广泛关注，这也预示着新闻行业智能化变革的进程将进一步加速。尤其是以ChatGPT为典型代表的AIGC技术，极有可能给整个新闻行业带来颠覆性变革。未来，AIGC技术的推广将改变新闻行业的基本运行逻辑，推动行业格局重构。它打破现有新闻生产规则，促使新闻产业链重塑。同时，新的新闻产品和业态不断涌现，推动新闻消费走向人性化。

本文深入探讨了文生视频大模型设计中的安全风险及其矫治策略。随着人工智能技术的快速发展，文生视频大模型如Sora和PixelDance等，已经能够根据文本描述生成视频内容，为影视、广告、教育等行业带来了革命性的变化。然而，这些技术进步也伴随着隐私泄露、数据安全、道德价值偏离等安全风险。本文分析了训练数据、提示词注入攻击、电信欺诈、道德价值偏离和人机交互等方面的风险，并介绍了差分隐私和联邦学习等风险治理策略。

在数字技术飞速迭代的当下，媒介已深度融入青少年的学习生活，成为其认知世界、社交互动的重要载体。然而，碎片化的媒介信息、多元的传播形态也给青少年带来了信息甄别困难、价值判断偏差、网络沉迷等诸多挑战，媒介素养的重要性愈发凸显。本文基于数字时代媒介传播的特征与青少年的认知发展规律，界定青少年媒介素养的核心内涵，结合江苏扬州、盐城两地5所中小学的问卷调查(有效问卷1086份)与深度访谈(100人)一手数据，剖析当前我国青少年媒介素养培养的现状与困境，探究困境形成的深层原因，引入平台社会理论、算法素养研究等前沿视角，针对生成式人工智能(AIGC)带来的新挑战补充素养培养内容，从家庭、学校、社会、媒介平台四方协同的视角，为不同年龄阶段、不同社会经济背景的青少年提出差异化培养路径，包括具体课程大纲、评价指标草案及典型案例剖析，为提升青少年媒介素养、引导青少年健康合理使用媒介提供理论参考与实践借鉴。

在社交媒体深刻重塑青年生活方式的背景下，本研究聚焦TikTok在马来西亚青年群体中的传播特征与社会效应。通过混合研究方法，分析平台内容生态如何适配青年兴趣偏好，探讨其在文化传播、社会动员等方面的积极影响，以及网络沉迷、信息失序、文化同质化等潜在挑战。研究发现，TikTok已成为马来西亚青年获取信息、表达文化认同的核心载体，既推动了本土传统艺术的全球化传播与青年创造力激发，也带来媒介依赖与价值观偏差等问题。研究进一步结合技术发展趋势与区域治理需求，从平台算法优化、政府监管创新、青年媒介素养提升三方面提出对策，旨在为构建平衡技术红利与社会风险的数字生态提供理论参考与实践路径。

近年来，人工智能在医疗方面的应用越来越广泛，2025年1月掀起了国产大模型DeepSeek的热潮，全国多家医院陆续接入了DeepSeek技术，如西安国际医学中心医院，该技术的接入可以提升诊断效率及就医体验。随着DeepSeek大模型在医疗领域广泛部署，其应用效果引发关注，但针对用户情感态度与使用意愿的系统研究仍然缺乏。为此，本研究采用Python爬虫从抖音平台采集4000余条评论，经数据清洗和分词处理，使用TF-IDF提取文本特征，构建Transformer情感分析模型，并运用LDA模型提取评论主题。结果表明，评论情感以正面为主，多数用户认可“DeepSeek + 医疗”的诊断准确性与便利性，但少量负面评论涉及隐私、安全及AI可靠性顾虑。LDA分析识别出智能医疗、就医服务、医疗变革、诊疗方案四个主要主题。据此提出提高诊断精准度、优化用户界面和加强隐私保护等建议，以提升用户信任度和使用意愿。

近年来大学生心理健康问题日益凸显，传统心理健康评估与干预方法的局限性以及人工智能技术的快速发展，为大学生心理健康评估与干预带来了新的发展机遇。文章探讨了人工智能技术在大学生心理健康评估和干预中用到的多模态评估技术、机器学习和神经网络算法、虚拟现实(VR)和增强现实(AR)技术、分析了“北小六”人工智能心理服务机器人、AI倾诉师EmoGPT人工智能聊天机器人、Woebot聊天机器人三个典型应用案例，讨论了在实际应用中可能面临的信息安全和伦理问题、缺乏共情能力的问题及应对措施，为将来人工智能技术在大学生心理健康服务领域的实践与研究提供参考。

本研究以337名山东某高校大学生为研究对象，探讨了交互质量与用户接受度之间的关系，以及心理距离和信任在两者间的链式中介作用。研究发现：交互质量能够显著正向预测接受度；交互质量能够通过心理距离和信任的链式中介作用间接影响接受度。本研究揭示了交互质量和接受度之间的内在机制，为人工智能在心理咨询领域的应用提供了理论基础。

近年来，人工智能技术正逐步渗透心理健康领域的各类场景并重塑该领域的形态与维度，显著提升了服务的可达性和响应效率。然而，由于数据质量、用户接受、隐私保护及从业者态度等方面因素，人工智能在赋能心理健康服务中的信任问题日益凸显，成为不容忽视的挑战。这种信任缺失不仅可能干扰人机协作，还可能影响心理健康服务的整体效果及患者的治疗依从性。为重建信任，本文提出构建负责任的人工智能、提升数据质量与多样性、增强信息透明度、保障用户隐私安全以及提高从业人员的信任和接受度等策略，以期推动人工智能技术在心理健康服务中的可信赖性，从而提升个体心理健康和社会福祉。

人工智能生成内容技术已深度融入教育领域中，为了探究影响信息专业类的学生应用AIGC技术的关键因素的作用机理，以期为推动“人工智能 + 教育”领域提供理论借鉴和实践指导，将技术接受模型和期望失验理论整合，结合感知有用性、感知易用性和满意度等构念构建AIGC运用概念模型，提出研究假设和设计调查问卷，采用偏最小二乘法分析数据并对测量模型和结构模型进行评估。研究发现：信息质量和感知绩效正向影响满意度，进而满意度、感知有用性和感知易用性正向影响AIGC工具的使用意愿。本研究除了揭示影响学生应用AIGC技术的关键因素及其作用机理以外，还结合研究结果给出教学实践启示，为数智时代的信息类专业的教学提供参考。

Background With the overwhelming availability of online health information and high prevalence of health misinformation, it is vital to understand the status and key influencing factors of its use among individuals. This study aims to explore the online health information-seeking behavior and preference of the influencing factors among college students. Methods We used the best-worst scaling approach to determine college students’ preferences for factors influencing online health information-seeking behavior. A total of 11 attributes of online health information seeking were confirmed by literature review and focus group, and a balanced incomplete block design was used to create 11 tasks for the BWS survey. An online survey was conducted from March 2023 to May 2023 using the BWS survey questionnaire. Results Both the BWS score and mixed logit model results indicate that “verified by professional institutions or health professionals”(mean BW=1.938; coefficient = 3.096), “information source from trustworthy and authoritative website”(mean BW = 1.921; coefficient = 3.015), “privacy and security guaranteed”(mean BW = 1.234; coefficient = 2.637), and “consistency of information” (mean BW = 0.803; coefficient = 2.313) were the most important factors and were valued more positively than negatively by respondents. The results showed the covariate of medical education had positive effects of 0.410 and 0.279 on the preference of “writing and language” and “professional interface design,” while medical education background had negative effects of −0.307 on the preference of “disclosure of author information.” Conclusion We recommend that concerned authorities consider interventions targeting the accuracy, credibility, privacy, and consistency of online health information management for college students.

Abstract Objective The current study examined how college students search online for mental health information and the impact of these searches on mental health literacy, stigma, and help-seeking. Method Undergraduate participants (N = 270; Fall 2015 to Spring 2019) were randomly assigned to search online for information about coping with anxiety for themselves or a friend (experimental activity), or to utilize Google Maps to answer navigational questions (control). Results Participants who conducted an online search demonstrated greater mental health literacy including optimism about psychotherapy, and lower levels of certain types of stigma, but lower willingness to seek/recommend professional help. Participants were more likely to recommend professional help for a friend compared to themselves. Conclusions Online searches for mental health information have the potential to increase mental health literacy. Universities can harness the Internet to reduce help-seeking barriers but should also address when it is appropriate to engage in self-help versus seek professional help.

Alternative health information seeking is an important way for older adults to access health support. In this study, we took college students who had conducted online health information substitution search for the elderly at home as the research subjects, and used the extended Comprehensive Model of Information Seeking (CMIS) as the theoretical framework to explore the behavioral characteristics and influencing factors of college students' online health information substitution search for the elderly at home through the questionnaire method (N=321), with a view to providing new ideas for the dissemination of health information to the elderly population. The study found that the content and search platforms of college students' alternative health information searching for the elderly at home are relatively basic; the influence of health awareness pressure on alternative health information searching behavior is not significant, and there may be other potential emotional mediators as well as insufficient behavioral internal motivation; filial piety culture influences the college students' alternative health information searching behavior through the utility of the information carrier; self-efficacy is an intrinsic motivation for intergenerational feedback; and the influence of self-efficacy is a key factor in college students' alternative health information searching behavior. Self-efficacy, as an intrinsic motivation for intergenerational feedback, is also an important factor influencing online health information alternative search behavior.

Background: Mental health problems are prevalent in college students. The COVID-19 pandemic exacerbated the problems, and created a surge in the popularity of telehealth and mobile health solutions. Despite that mobile health is a promising approach to help students with mental health needs, few studies exist in investigating key features students need in a mental health self-management tool. Objective: The objective of our study was to identified key requirements and features for the design of a student-centered mental health self-management tool. Methods: An interview study was first conducted to understand college students' needs and preferences on a mental health self-management tool. Functional information requirement analysis was then conducted to translate the needs into design implications. Results: A total of 153 university students were recruited for the semi-structured interview. The participants mentioned several features including coping techniques, artificial intelligence, time management, tracking, and communication with others. Participant's preferences on usability and privacy settings were also collected. The desired functions were analyzed and turned into design-agnostic information requirements. Conclusions: This study documents findings from interviews with university students to understand their needs and preferences for a tool to help with self-management of mental health.

No abstract available

Mobile application (app), with its expansive reservoir of data and content, harbors considerable promise in advancing health outcomes at both the individual and population levels. Nevertheless, there is a paucity of scholars that concretely examine the dynamics of health information acquisition within mobile app. This research presents a conceptual model aimed at investigating the potential ramifications of health information acquisition on both individuals' psychological and physical well-being. Concurrently, this research attempts to illuminate the underlying mechanisms behind these correlations through evaluating the mediating role of bonding and bridging social capital. The obtained results of a web-based survey conducted among 656 college students in mainland China suggest a positive association between health information acquisition and bonding and bridging social capital. Moreover, the study reveals that the impact of health information acquisition via mobile applications on psychological and physical well-being is significantly mediated by college students' bonding and bridging social capital. The cultivation of bonding social capital could exert a direct and positive influence on college students' physical well-being. However, there appears to be no discernible correlation between bridging social capital and physical well-being. Taken collectively, these findings not only complement extant theoretical perspectives within the scholarship concerning mobile app usage for health improvement, but also furnish several pragmatic guidelines for healthcare professionals and mobile app designers.

The current phenomenon that has become a new trend among young people is the circulation of a web page that discusses mental health tests whose sources are unknown. This makes students self-diagnose the information they read and obtain on social media so that various views on mental health arise. The direction of this research is to understand the views of STIKes KHG students about mental health self-diagnosis. The technique used in this research is descriptive qualitative. The sample was obtained using purposive sampling technique as many as five informants, data obtained from in-depth questions and answers with respondents. The research obtained six themes, namely: 1) Knowledge and understanding of STIKes KHG students about self-diagnosis of mental health 2) Experience of STIKes KHG students about self diagnosis in mental health 3) Coping mechanisms carried out by STIKes KHG students if they experience self diagnosis, 4) Sources known to STIKes KHG students about self diagnosis 5) Positive impact of the internet or social media for STIKes KHG students 6) Negative impact of the internet or social media for STIKes KHG students. Based on the results of this research, it can be concluded that the majority of STIKes KHG students know and have experience about mental health self-diagnosis from the internet or social media so that various perceptions arise based on information obtained independently.

In the conditions of the rapid development of Information Technologies, online culture has firmly entered the life of students. This article discusses the impact of internet culture on students ' mental health and the temporary effect of Information Literacy. Hierarchical random sampling selected 1,200 students from five domestic universities to conduct the questionnaire. The scale covers many aspects such as online cultural contacts, information literacy, and mental health. The collected data are analyzed using the model of structural equations and the method of regression analysis. Data shows that online culture will have a significant impact on students ' mental health, and its positive and negative effects coexist. High-quality educational resources and positive social interaction online are positive factors that can improve mental health; while harmful information shocks, such as rumors and excessive fun, can easily aggravate negative emotions and cause an increased risk of symptoms, including anxiety or depression. Overall, the greater ability to recognize and apply information helps to maintain a more optimistic state of mental health through the effective identification and use of positive network resources and the removal of unwanted disruptions in information, and therefore finds that information literacy plays an important role in regulating the above relationships. This study not only expands the cognitive depth of the relationship model between online culture and health psychology, but provides a new direction and enlightening value for effective ways to use enhanced information opportunities to solve potential problems at a practical level of Education.

Objective Generative AI is increasingly used to provide health-related information in addition to online health information seeking (OHIS). Users’ willingness to adopt it is crucial. This study investigates individual factors associated with more frequent OHIS: health status, health anxiety, and eHealth literacy. Using the Technology Acceptance Model (TAM), we examined whether these factors are related to more trust in generative AI for health-related purposes and the willingness to use it. Methods Using SEM, we analyzed cross-sectional survey data (N = 4775) that is representative of adult Czech internet users (50% female; aged 18–95 years). Results Trust in AI was strongly associated with willingness to use AI. Health status and health anxiety were related to willingness to use AI only indirectly through trust. Higher eHealth literacy was associated with more trust only marginally and had no direct relationship with willingness to use AI. Wellness-related OHIS was positively associated with willingness to use AI for wellness purposes, and illness-related OHIS was associated with willingness to use AI for illness purposes. Conclusion Although not emphasized in TAM and its health-related extensions, trust seems to be a critical mediator in the adoption of generative AI for health purposes. Other factors related to OHIS were not associated with willingness to use AI, except for their relationship with trust. eHealth literacy is practically unrelated to trust and willingness to use AI, which is noteworthy given that health anxiety and health status related to higher acceptance are associated with more risky or high-stake use of online health information.

The deployment of Conversational User Interfaces (CUIs) with advanced Large Language Models (LLMs) has significantly transformed health information seeking and dissemination, facilitating immediate and interactive communication between users and digital health resources. However, while trust is crucial for adopting health advice, how the dissemination interface influences people's perceived trust in health information provided by LLMs remains unclear. To address this, we conducted a mixed-methods, within-subjects lab study (N=20) to investigate how different CUIs (i.e., a text-based, speech-based, and embodied interface) affect user-perceived trust levels when delivering health information from an identical LLM source. Our key findings showed that: (a) participants' trust levels in health information delivered were significantly variant across different interfaces; (b) there are significant correlations between trust in health-related information and trust in the delivered interface as well as the usability level of the interface; (c) the type of health questions did not affect participants' perceived trust. Besides, we identified key factors influencing trust in health information delivered through various CUIs and explored differences in how people trust health information from LLM and its dissemination. We highlight the potential of LLM-powered CUIs in supporting health-related information-seeking behaviors. This work contributes insights for ensuring effective and trustworthy personal health information-seeking in the era of LLM-powered CUIs and multi-modal information dissemination.

People increasingly rely on online sources for health information seeking due to their convenience and timeliness, traditionally using search engines like Google as the primary search agent. Recently, the emergence of generative Artificial Intelligence (AI) has made Large Language Model (LLM) powered conversational agents such as ChatGPT a viable alternative for health information search. However, while trust is crucial for adopting the online health advice, the factors influencing people's trust judgments in health information provided by LLM-powered conversational agents remain unclear. To address this, we conducted a mixed-methods, within-subjects lab study (N=21) to explore how interactions with different agents (ChatGPT vs. Google) across three health search tasks influence participants' trust judgments of the search results as well as the search agents themselves. Our key findings showed that: (a) participants' trust levels in ChatGPT were significantly higher than Google in the context of health information seeking; (b) there is a significant correlation between trust in health-related information and trust in the search agent, however only for Google; (c) the type of search tasks did not affect participants' perceived trust; and (d) participants' prior knowledge, the style of information presentation, and the interactive manner of using search agents were key determinants of trust in the health-related information. Our study taps into differences in trust perceptions when using traditional search engines compared to LLM-powered conversational agents. We highlight the potential role LLMs play in health-related information-seeking contexts, where they excel as stepping stones for further search. We contribute key factors and considerations for ensuring effective and reliable personal health information seeking in the age of generative AI.

RATIONALES Using AI for health information seeking is a novel behavior, and as such, developing effective communication strategies to optimize AI adoption in this area presents challenges. To lay the groundwork, research is needed to map out users' behavioral underpinnings regarding AI use, as understanding users' needs, concerns and perspectives could inform the design of targeted and effective communication strategies in this context. OBJECTIVE Guided by the planned risk information seeking model and the comprehensive model of information seeking, our study examines how socio-psychological factors (i.e., attitudes, perceived descriptive and injunctive norms, self-efficacy, technological anxiety) and factors related to information carriers (i.e., trust in and perceived accuracy of AI), shape users' latent profiles. In addition, we explore how individual differences in demographic attributes and anthropocentrism predict membership in these user profiles. METHODS We conducted a quota-sampled survey with 1051 AI-experienced users in Hong Kong. Latent profile analysis was used to examine users' profile patterns. The hierarchical multiple logistic regression was employed to examine how individual differences predict membership in these user profiles. RESULTS The latent profile analysis revealed five heterogeneous profiles, which we labeled "Discreet Approachers," "Casual Investigators," "Apprehensive Moderates," "Apathetic Bystanders," and "Anxious Explorers." Each profile was associated with specific predictors related to individual differences in demographic attributes and/or aspects of anthropocentrism. CONCLUSION The findings advance theoretical understandings of using AI for health information seeking, provide theory-driven strategies to empower users to make well-informed decisions, and offer insights to optimize the adoption of AI technology.

Artificial intelligence (AI) integration in clinical practice has intensified in the last few years, from systems analysing and interpreting existing data to generative AI systems capable of creating new information and offering new possibilities for patient communication. However, the public’s perception of AI- generated health information remains largely unexplored. This study aimed to assess public trust in AI-generated health information, identifying influencing factors on their trust and evaluating the accuracy of AI-produced content. A mixed-method approach was employed, involving a survey distributed via social media to individuals with recent access to health information. Results revealed that while the public knew AI systems’ capabilities, their trust in AI-generated content was moderate. Key concerns included: the accuracy of the information, potential biases in AI algorithms, and ethical issues related to privacy. Results showed that transparency, healthcare professional endorsements, and clear evidence of accuracy are critical in building trust in AI-generated health information. Addressing these concerns is essential for successfully integrating AI into patient communication, to enable the reliability and use of AI as an ethical tool in healthcare.

Abstract The commercialization of artificial intelligence (AI) in healthcare is accelerating, yet academic research on its users remains scarce. To what extent are they willing to disclose personal health privacy to AI doctors compared to traditional human doctors? What factors are shaping these decisions? The lack of user research has left these questions unanswered. This article, based on privacy calculus theory, conducted a multi-factorial between-subjects online experiment (N = 582) with a 2 (medical provider: AI vs. human) × 2 (emotional support: low vs. high) × 2 (information sensitivity: low vs. high) design. The results indicated that AI doctors lead participants to perceive both lower health benefits and privacy risks. Emotional support is not always beneficial. On one hand, high emotional support can provide patients with more health benefits, but on the other hand, it also poses higher levels of privacy risks. Additionally, high emotional support responses from AI doctors could enhance patients’ health benefits, trust, and willingness to disclose health privacy, while the opposite was observed for human doctors.

Artificial Intelligence is changing the way people look for health and managing the health information. This study explores how perceived disease risk, health related stigma and personal motivation effects the use of AI health tools and how these factors influence the trust satisfaction and health outcomes. A questionnaire was used to collect data from different AI platform users, and convenience sampling was applied. Descriptive, correlation and reliability analysis was used to understand the data. The results shows that people are more likely to use AI Health tools when they find any ear or threat to their health, and those who faces social stigma prefer private and confidential AI health tools. Factors like fast response, low or no costs, and privacy play an important role in adopting AI health tools. Trust in each platform increases the user satisfaction, and positive health outcomes improves their overall success. The study shows that AI health tools can be helpful, accessible and safe resource, especially when stigma, limited access or social pressure makes traditional health care more difficult. These findings can help design AI health tools that are faster, user friendly, culture aware and supportive of better health decisions and wellbeing.

Today, TikTok is not just a social media platform for sharing trend-based videos, but it has evolved into a trusted source of information, particularly in areas such as health, parenting, and lifestyle. This shift is influenced by TikTok's algorithm and the way users interact with content. Through netnographic analysis and FGD, this study examines how TikTok has influenced the changing ways people build trust in the digital world, using heuristic trust theory. TikTok's AI algorithm shapes new forms of trust through several factors: reputation, endorsement, consistency, emotional relevance and popularity. The algorithm promotes high-engagement content & user interactions in the comment section. The findings reveal that in digital era, trust is formed instantly; relying more on pragmatic short cut than careful cognitive considerations. This makes digital literacy essential to ensure accurate understanding and evaluation of information. While this shift in how trust is formed brings positive impacts for certain parties, it also opens up the possibility for negative consequences to emerge.

Artificial Intelligence Generated Content (AIGC) is reshaping the ecology of social media and even online media with unprecedented authenticity and generation efficiency. As AI text-image and video generation technologies mature, AIGC visual news has achieved a deceptive level of authenticity, continuously blurring the line between fact and fiction. As a special information form relying on authenticity and public trust, the impact of AIGC visual news on the public trust system urgently requires in-depth exploration. Therefore, this study collects public evaluations and attitudes toward AIGC visual news from news platforms and social media, quantitatively analyzes public trust levels, and systematically explores its mechanism of action on the public trust system. The results show that AIGC visual news has triggered a significant trust deficit, whose crisis mainly stems from multiple factors such as the subversion of individual cognitive frameworks, the collapse of traditional trust foundations, and the low-cost abuse of technology. This influence has a two-sided nature, on the one hand, it weakens the traditional principles of trust; on the other hand, it compels the society to start the process of cognitive restructuring to facilitate the creation of a more reasonable and transparent trust mechanism.

Generation Z is the age group that has experienced the most mental health-related issues due to the COVID-19 pandemic in Indonesia. Mental health information system services have become increasingly popular, so patient services are overwhelmed. Chatbot has emerged as one of the solutions to address this problem. However, conversations with them have led to social issues alongside the growing use of chatbots. This study aims to identify the factors influencing Generation Z's trust in patient service chatbots within mental health applications in Indonesia. Using the Interpretative Phenomenological Analysis method, this research analyses qualitative data from observations and interviews with five undergraduate students. Based on the analysis, the study identified seven factors that influence Generation Z's trust in chatbot customer service within mental health applications, with three being novel findings.

Rumors, conspiracies, and health‐related misinformation have gone hand‐in‐hand with the global COVID‐19 pandemic, making it hard to obtain reliable and accurate information. Against this background, this study examined the different psychosocial predictors of believing in conspiratorial information related to general health in India. Indian participants (N = 826) responded to measures related to conspiratorial thinking, trust, moral emotions, political ideology, bullshit receptivity, and belief in conspiratorial information in an online survey. Exploratory and confirmatory factor analyses were used to determine the validity of the instruments used with an Indian sample. Results revealed that lower subjective socioeconomic status, lower trust in political institutions, greater negative moral emotions, greater conspiratorial thinking, and right‐leaning political ideology predicted beliefs in health‐related conspiratorial information. In highlighting these potential psychosocial determinants of conspiratorial beliefs, we can move toward combating conspiracies effectively and developing necessary interventions for the same. Future work can focus on assessing the moderating effects of political ideology on conspiratorial beliefs in India.

Online Mental Health Communities (OMHCs) are becoming vital spaces for young people to access peer-based mental health support. However, the behavioral factors that influence continued participation in these communities remain underexplored. This study extends the Theory of Planned Behavior (TPB) by including initial trust, OMHC engagement, emotional support, informational support, and perceived anonymity to examine young users’ intentions to sustain their participation in OMHCs. Using Partial Least Squares Structural Equation Modeling (PLS-SEM), data from 459 Filipino youth aged 18 to 30 were analyzed to test the model. Results revealed that attitude, subjective norms, and perceived behavioral control significantly predict sustainable use intentions. Emotional and informational support strongly influenced subjective norms, while initial trust and OMHC engagement influenced attitude. Perceived anonymity, however, did not exhibit a significant effect. The findings suggest that trust, peer support, and user engagement play an important role in shaping participation in digital mental health spaces. The article concludes with practical implications for platform developers, mental health professionals, and policymakers who aim to improve access to mental health support online. By addressing the psychosocial dynamics shaping OMHC participation, the research advances the understanding of help-seeking behaviors in low-resource and collectivist settings. Plain Language Summary How trust and support affect young people’s involvement in online mental health communities Access to mental health resources is limited worldwide, especially for young adults in developing countries like the Philippines. This study examines what influences Filipino young adults to join and stay active in online mental health communities (OMHCs). Researchers surveyed 459 participants aged 18-30 and used a statistical model to find patterns in their behavior. We discovered that young people's continued use of OMHCs is influenced by their attitudes, the support they feel from others, and how much control they believe they have over their participation. Trust in the platform and feeling emotionally supported also helped shape positive attitudes. Interestingly, staying anonymous was not as important as expected. The study shows that feeling supported, engaged, and trusting the platform matters more than hiding one’s identity. These insights can help improve the design of mental health platforms and guide professionals and policymakers in making digital support more accessible, especially in places with limited mental health resources like the Philippines.

Aim of the Study: The study aimed to explore the impact of social media influencers on adults’ health behavior, with a particular focus on the mediating roles of credibility, trust, and emotional resonance. Drawing on Parasocial Interaction Theory, the research sought to understand not only whether influencers shape health practices, but also how emotional and relational factors contribute to this influence. Methodology: A survey of 245 adults was conducted to assess their exposure to influencers, perceptions of credibility and trust, emotional resonance, and self-reported health behaviors. Data was collected using closes ended questionnaire using convenient sampling technique in Rawalpindi, Pakistan. Data were analyzed using multiple hierarchical regression, ANOVA, independent-samples t-tests, and structural equation modeling to test both direct and indirect pathways. Findings: The results showed that social media influencers significantly predicted health behavior both directly and indirectly. Emotional resonance emerged as the strongest mediator, exerting a large positive effect, while credibility had a small negative influence and trust was not significant. Participants with higher exposure to influencers reported healthier behaviors than those with lower exposure, confirming the role of social media engagement in shaping lifestyle choices. The SEM model demonstrated partial mediation, with influencers continuing to exert a strong direct effect even after accounting for mediators. Conclusion: The study concludes that the effectiveness of influencers lies not only in the accuracy of the information they provide but also in the emotional connections they establish with their audiences. These findings extend Parasocial Interaction Theory into digital health contexts and highlight the importance of authenticity and relatability in persuasive communication. For practice, the results suggest that public health campaigns can benefit from collaborating with influencers who are both emotionally engaging and ethically responsible.

This study investigates the intricate relationship between exposure to information sources, trust in these sources, conspiracy and misinformation beliefs, and COVID-19 anxiety among 509 Omani citizens aged 11 to 50, representing 11 governorates. Employing structural equation modeling, we not only examine these associations but also explore how trust and COVID-19 anxiety act as moderating variables in this context. Additionally, we delve into demographic factors such as age group, educational level, gender, and place of residence (governorate) to discern potential variations. Our findings reveal that trust in health experts is inversely related to belief in conspiracy theories, while trust in health experts negatively correlates with exposure to conspiracy and misinformation. Intriguingly, trust in health experts exhibits divergent effects across governorates: it diminishes conspiracy and misinformation beliefs in some regions but not in others. Exposure to personal contacts and digital media, on the other hand, is associated with heightened beliefs in misinformation and conspiracy theories, respectively, in select governorates. These distinctions may be attributed to proximity to Muscat, the capital city of Oman, where various media outlets and policy-making institutions are situated. Furthermore, lower educational attainment is linked to greater belief in conspiracy and misinformation. Females reported higher levels of conspiracy theory beliefs and COVID-19 anxiety while no significant differences were detected in misinformation beliefs. This study sheds light on the intricate dynamics of misinformation and conspiracy theories in the context of COVID-19 in Oman, highlighting the pivotal roles of trust and COVID-19 anxiety as moderating factors. These findings offer valuable insights into understanding and addressing the spread of misinformation and conspiracy theories during a public health crisis.

Background The last decade has witnessed the rapid development of health care conversational agents (CAs); however, there are still great challenges in making health care CAs trustworthy and acceptable to patients. Objective Focusing on intelligent guidance CAs, a type of health care CA for web-based patient triage, this study aims to investigate how anthropomorphic cues influence patients’ perceived anthropomorphism and social presence of such CAs and evaluate how these perceptions facilitate their trust-building process and acceptance behavior. Methods To test the research hypotheses, the video vignette methodology was used to evaluate patients’ perceptions and acceptance of various intelligent guidance CAs. The anthropomorphic cues of CAs were manipulated in a 3×2 within-subject factorial experiment with 103 participants, with the factors of agent appearance (high, medium, and low anthropomorphic levels) and verbal cues (humanlike and machine-like verbal cues) as the within-subject variables. Results The 2-way repeated measures ANOVA analysis indicated that the higher anthropomorphic level of agent appearance significantly increased mindful anthropomorphism (high level>medium level: 4.57 vs 4.27; P=.01; high level>low level: 4.57 vs 4.04; P<.001; medium level>low level: 4.27 vs 4.04; P=.04), mindless anthropomorphism (high level>medium level: 5.39 vs 5.01; P<.001; high level>low level: 5.39 vs 4.85; P<.001), and social presence (high level>medium level: 5.19 vs 4.83; P<.001; high level>low level: 5.19 vs 4.72; P<.001), and the higher anthropomorphic level of verbal cues significantly increased mindful anthropomorphism (4.83 vs 3.76; P<.001), mindless anthropomorphism (5.60 vs 4.57; P<.001), and social presence (5.41 vs 4.41; P<.001). Meanwhile, a significant interaction between agent appearance and verbal cues (.004) was revealed. Second, the partial least squares results indicated that privacy concerns were negatively influenced by social presence (β=−.375; t312=4.494) and mindful anthropomorphism (β=−.112; t312=1.970). Privacy concerns (β=−.273; t312=9.558), social presence (β=.265; t312=4.314), and mindless anthropomorphism (β=.405; t312=7.145) predicted the trust in CAs, which further promoted the intention to disclose information (β=.675; t312=21.163), the intention to continuously use CAs (β=.190; t312=4.874), and satisfaction (β=.818; t312=46.783). Conclusions The findings show that a high anthropomorphic level of agent appearance and verbal cues could improve the perceptions of mindful anthropomorphism and mindless anthropomorphism as well as social presence. Furthermore, mindless anthropomorphism and social presence significantly promoted patients’ trust in CAs, and mindful anthropomorphism and social presence decreased privacy concerns. It is also worth noting that trust was an important antecedent and determinant of patients’ acceptance of CAs, including their satisfaction, intention to disclose information, and intention to continuously use CAs.

Cyberchondria, characterized by excessive online health information seeking and resulting anxiety, is intensifying. This study aims to examine how threat perceptions and cognitive factors drive cyberchondria and how this condition leads to health information fatigue on social media (HIFSM), self-medication and therapy compliance. This study integrates protection motivation theory, cognitive load theory and the stressor-strain-outcome (SSO) model to inform the partial least squares path modeling of a 2-wave survey over 6 months of 400 participants. Perceived susceptibility, perceived severity, online information trust and information overload intensify cyberchondria, which sparks HIFSM and, in turn, increases self-medication while undermining therapy compliance. Trust in physicians mitigates these adverse effects. Since information overload fuels cyberchondria, the findings urge social media developers to help curb cyberchondria by prioritizing credible health content, integrating source-verification features and collaborating with clinicians to curate guideline-based information. This study advances cyberchondria research by uniting three theoretical perspectives and identifying physician trust as a protective factor.

Abstract Background Disinformation on social media can seriously affect mental health by spreading false information, increasing anxiety, stress, and confusion in vulnerable individuals, as well as perpetuating stigma. This flood of misleading content can undermine trust in reliable sources and heighten feelings of isolation and helplessness among users. Objective This study aimed to explore the phenomenon of disinformation about mental health on social media and provide recommendations to mental health professionals that would use social media platforms to create educational videos about mental health topics. Methods A comprehensive analysis conducted on 1000 TikTok videos from more than 16 countries, available in English, French, and Spanish, covering 26 mental health topics. The data collection was conducted using a framework on disinformation and social media. A multilayered perceptron algorithm was used to identify factors predicting disinformation. Recommendations to health professionals about the creation of informative mental health videos were designed as per the data collected. Results Disinformation was predominantly found in videos about neurodevelopment, mental health, personality disorders, suicide, psychotic disorders, and treatment. A machine learning model identified weak predictors of disinformation, such as an initial perceived intent to disinform and content aimed at the general public rather than a specific audience. Other factors, including content presented by licensed professionals such as a counseling resident, an ear-nose-throat surgeon, or a therapist, and country-specific variables from Ireland, Colombia, and the Philippines, as well as topics such as adjustment disorder, addiction, eating disorders, and impulse control disorders, showed a weak negative association with disinformation. In terms of engagement, only the number of favorites was significantly associated with a reduction in disinformation. Five recommendations were made to enhance the quality of educational videos about mental health on social media platforms. Conclusions This study is the first to provide specific, data-driven recommendations to mental health providers globally, addressing the current state of disinformation on social media. Further research is needed to assess the implementation of these recommendations by health professionals, their impact on patient health, and the quality of mental health information on social networks.

Abstract Background Generative artificial intelligence (GenAI) models have emerged as a promising yet controversial tool for mental health. Objective The purpose of this study is to understand the experiences of individuals who repeatedly used ChatGPT (GenAI) for emotional and mental health support (EMS). Methods We recruited 270 adult participants across 29 countries who regularly used ChatGPT (OpenAI) for EMS during April 2024. Participants responded to quantitative survey questions on the frequency and helpfulness of using ChatGPT for EMS, and qualitative questions regarding their therapeutic purposes, emotional experiences of using, and perceived helpfulness and rationales. Thematic analysis was used to analyze qualitative data. Results Most participants reported using ChatGPT for EMS at least 1‐2 times per month for purposes spanning traditional mental health needs (diagnosis, treatment, and psychoeducation) and general psychosocial needs (companionship, relational guidance, well-being improvement, and decision-making). Users reported various emotional experiences during and after use for EMS (eg, connected, relieved, curious, embarrassed, or disappointed). Almost all users found it at least somewhat helpful. The rationales for perceived helpfulness include perceived changes after use, emotional support, professionalism, information quality, and free expression, whereas the unhelpful aspects include superficial emotional engagement, limited information quality, and lack of professionalism. Conclusion Despite the absence of ethical regulations for EMS use, GenAI is becoming an increasingly popular self-help tool for emotional and mental health support. These results highlight the blurring boundary between formal mental health care and informal self-help and underscore the importance of understanding the relational and emotional dynamics of human-GenAI interaction. There is an urgent need to promote AI literacy and ethical awareness among community users and health care providers and to clarify the conditions under which GenAI use for mental health promotes well-being or poses risk.

The rapid advancement of Large Language Models has sparked heated debate over whether Generative Artificial Intelligence (AI) chatbots can serve as “digital therapists” capable of providing therapeutic support. While much of this discussion focuses on AI’s lack of agency, understood as the absence of mental states, consciousness, autonomy, and intentionality, empirical research on users’ real-world experiences remains limited. This study explores how individuals with mental distress experience support from both generative AI chatbots and human psychotherapy in natural and unguided contexts, with a focus on how perceptions of agency shape therapeutic experiences. By drawing on participants’ dual exposure, the study seeks to contribute to the ongoing debate about “AI therapists” by clarifying the role of agency in therapeutic change. Sixteen adults who had sought mental health support from both human therapists and ChatGPT participated in semi-structured interviews, during which they shared and compared their experiences with each type of interaction. Transcripts were analyzed using reflexive thematic analysis. Three themes captured participants’ perceptions of ChatGPT relative to human therapists: (1) encouraging open and authentic self-disclosure but limiting deep exploration; (2) the myth of relationship: caring, acceptance, and understanding; (3) fostering therapeutic change: the promise and pitfalls of data-driven solutions. We propose a conceptual model that illustrates how differences in agency status between AI chatbots and human therapists shape the distinct ways they support individuals with mental distress, with agency functioning as both a strength and a limitation for therapeutic engagement. Given that agency functions as a double-edged sword in therapeutic interactions, future mental health services should consider integrated care models that combine the non-agential advantages of AI chatbots with the agentic qualities of human therapists. Rather than anthropomorphizing AI chatbots, their non-agential features—such as responsiveness, absence of intentions, objectivity, and disembodiment—should be strategically leveraged to complement specific functions in human-delivered psychotherapy. At the same time, practitioners should maximize the benefits of their agentic qualities while remaining cautious of the risks. The findings should be interpreted with caution as the sample consisted mainly of young, well-educated Chinese participants from a collectivist cultural context, which may limit transferability to other populations, particularly those from individualistic cultures with different mental health literacy levels, stigma patterns, and therapeutic norms. Not applicable.

Generative artificial intelligence (GenAI) is rapidly emerging as a powerful intermediary for information access, reshaping how individuals seek and evaluate knowledge. While prior research has examined how people use conversational GenAI to find specific types of information and how dialogue-based search compares with traditional, non-dialogue search engines such as Google, less is known about the psychological and perceptual antecedents of this behavior. Much of the existing literature emphasizes rational factors, such as perceived information-gathering capabilities, yet affective dimensions like emotional attachment have received limited attention. Recent studies suggest that users who form emotional attachment to ChatGPT tend to rely on it for information and emotional support. Using a national US sample (N = 900) collected by Verasight, this study examines how emotional attachment to GenAI, trust in AI, AI literacy, and risk perception influence individuals’ GenAI usage to seek various types of information. Results show that emotional attachment to GenAI is significantly associated with higher trust in GenAI, which in turn is associated with using it as an information intermediary across various domains. Additionally, the association between emotional attachment and trust in GenAI is more pronounced among individuals with low levels of AI literacy and those who perceive high risk from using GenAI for information access. In addition, AI literacy moderated the relationship between emotional attachment and health-related and political information seeking through GenAI. Findings contribute to a deeper understanding of the emotional dynamics shaping human–AI interaction and offer practical insights for the design and governance of AI systems in public information environments.

Objective Generative artificial intelligence (genAI) has become popular for the general public to address mental health needs despite the lack of regulatory oversight. Our study used a digital ethnographic approach to understand the perspectives of individuals who engaged with a genAI tool, ChatGPT, for psychotherapeutic purposes. Methods We systematically collected and analyzed all Reddit posts from January 2024 containing the keywords “ChatGPT” and “therapy” in English. Using thematic analysis, we examined users’ therapeutic intentions, patterns of engagement, and perceptions of both the appealing and unappealing aspects of using ChatGPT for mental health needs. Results Our findings showed that users utilized ChatGPT to manage mental health problems, seek self-discovery, obtain companionship, and gain mental health literacy. Engagement patterns included using ChatGPT to simulate a therapist, coaching its responses, seeking guidance, re-enacting distressing events, externalizing thoughts, assisting real-life therapy, and disclosing personal secrets. Users found ChatGPT appealing due to perceived therapist-like qualities (e.g. emotional support, accurate understanding, and constructive feedback) and machine-like benefits (e.g. constant availability, expansive cognitive capacity, lack of negative reactions, and perceived objectivity). Concerns regarding privacy, emotional depth, and long-term growth were raised but rather infrequently. Conclusion Our findings highlighted how users exercised agency to co-create digital therapeutic spaces with genAI for mental health needs. Users developed varied internal representations of genAI, suggesting the tendency to cultivate mental relationships during the self-help process. The positive, and sometimes idealized, perceptions of genAI as objective, empathic, effective, and free from negativity pointed to both its therapeutic potential and risks that call for AI literacy and increased ethical awareness among the general public. We conclude with several research, clinical, ethical, and policy recommendations.

PurposeThis study examined how people assess health information from AI and improve their diagnostic ability to identify health misinformation. The proposed model was designed to test a cognitive heuristic theory in misinformation discernment.Design/methodology/approachWe proposed the heuristic-systematic model to assess health misinformation processing in the algorithmic context. Using the Analysis of Moment Structure (AMOS) 26 software, we tested fairness/transparency/accountability (FAccT) as constructs that influence the heuristic evaluation and systematic discernment of misinformation by users. To test moderating and mediating effects, PROCESS Macro Model 4 was used.FindingsThe effect of AI-generated misinformation on people’s perceptions of the veracity of health information may differ according to whether they process misinformation heuristically or systematically. Heuristic processing is significantly associated with the diagnosticity of misinformation. There is a greater chance that misinformation will be correctly diagnosed and checked, if misinformation aligns with users’ heuristics or is validated by the diagnosticity they perceive.Research limitations/implicationsWhen exposed to misinformation through algorithmic recommendations, users’ perceived diagnosticity of misinformation can be predicted accurately from their understanding of normative values. This perceived diagnosticity would then positively influence the accuracy and credibility of the misinformation.Practical implicationsPerceived diagnosticity exerts a key role in fostering misinformation literacy, implying that improving people’s perceptions of misinformation and AI features is an efficient way to change their misinformation behavior.Social implicationsAlthough there is broad agreement on the need to control and combat health misinformation, the magnitude of this problem remains unknown. It is essential to understand both users’ cognitive processes when it comes to identifying health misinformation and the diffusion mechanism from which such misinformation is framed and subsequently spread.Originality/valueThe mechanisms through which users process and spread misinformation have remained open-ended questions. This study provides theoretical insights and relevant recommendations that can make users and firms/institutions alike more resilient in protecting themselves from the detrimental impact of misinformation.Peer reviewThe peer review history for this article is available at: https://publons.com/publon/10.1108/OIR-04-2023-0167

Background: Depression and anxiety affect nearly 1 in 4 Canadians. Traditional patient education materials, such as handouts, are often lengthy and difficult to understand, leading to disengagement. Human-like artificial intelligence (AI) avatars offer a novel way to supplement education by delivering consistent, engaging video content that mimics human interaction and is easily accessible online. Objective: This pilot study aimed to develop a human-like, non-generative AI avatar educational video to support education on antidepressants for patients living with depression and anxiety. The secondary objectives were to evaluate participants perceptions of the tool across 3 domains: credibility, satisfaction, and understanding. Methods: The video was developed through 2 Plan-Do-Study-Act (PDSA) cycles, informed by prior research on patient-reported barriers and enablers to antidepressant use. After viewing the video, participants completed a survey assessing the 3 domains. Success was predefined as ≥60% of participants rating each domain ≥4 on a 5-point Likert scale. Open-ended feedback was summarized descriptively to help inform revisions. Results: Fifteen University Health Network (UHN) Patient Partners participated in PDSA Cycle 1, most with lived experience of depression or anxiety and high digital literacy. Success thresholds were achieved for credibility (75%) and satisfaction (67%) but not for understanding (50%). After revisions, 10 participants from the original group completed PDSA Cycle 2, where all domains exceeded thresholds (credibility 90%, satisfaction 85%, understanding 82%). Participants described the tool as trustworthy, clear, and engaging. Conclusion: This pilot study demonstrated that human-like, non-generative AI avatars can be an effective supplementary educational tool to deliver education on antidepressants for individuals with depression and anxiety. The tool demonstrated acceptability across credibility, satisfaction, and perceived understanding, highlighting its potential to enhance patient engagement and access to reliable information. As a scalable and adaptable format, avatar-based education may extend beyond mental health to other conditions, languages, and clinical settings. Future studies should examine its impact on knowledge retention, treatment adherence, and integration into clinical practice.

Artificial intelligence (AI) technologies may become a dangerous means of psychological and socio-political destabilization in the hands of malicious actors. This article provides a comprehensive analysis of the key threats associated with the phenomenon of malicious use of AI (MUAI). The high pace of Egypt's industrial, scientific and technological development, combined with acute economic, social contradictions, political challenges, create favorable conditions for the growth of MUAI threats in this country. A three-level approach is used to assess the MUAI threats and their impact on psychological security. The first level of threats is characterized by the spread of deliberately distorted information about the nature and capabilities of AI. The second level pertains to cases of MUAI where public consciousness is not the direct target, but negative consequences occur nonetheless—due to harm inflicted on critical infrastructure, public health systems, energy and transport security, as well as individuals’ lives, health, and property. At the third and highest level of threat, AI is employed as a tool for manipulative influence on public consciousness with the explicit aim of social destabilization. Among the main MUAI threats in Egypt are: manipulation of alarming narratives about technological unemployment (the first level), growth of digital fraud with the increasing use of AI technologies (the second level), use of deepfake technologies and generative AI to create and spread disinformation (the third level). The article emphasizes the need to develop comprehensive strategies to counter MUAI, including strengthening regulatory frameworks, enhancing digital literacy, and advancing national cybersecurity mechanisms.

The quality of psychological nursing care for patients with mood disorders (MD) largely depends on caregivers' emotional assessment skills, communication abilities, and crisis identification capabilities. However, traditional psychological nursing education models exhibit significant limitations in resource allocation, practical scenario development, and individualized training. With the rapid advancement of Artificial Intelligence Generated Content (AIGC) technology, it demonstrates unique advantages in simulating emotional scenarios, generating diverse patient feedback, and providing real-time instructional support. Although AIGC applications in medical education are expanding, systematic research remains lacking regarding its development pathways, effective mechanisms, and implementation strategies in psychological nursing education informatization. To establish an AIGC-assisted framework for digitalized psychological nursing education and explore its digital development pathways for cultivating nursing competencies in emotional disorders, this study employed a mixed-method approach comprising expert interviews (n = 18), scenario-based teaching experiments (involving 162 nursing students), and system requirements analysis. The AIGC system generated emotion fluctuation cases, crisis dialogue scripts, and personalized learning feedback. Participants were randomly assigned to an experimental group (n = 82) or control group (n = 80) based on class assignments. Both groups showed no statistically significant differences in baseline characteristics including gender, age, and prior course foundations, ensuring comparable teaching interventions. The experimental group utilized an AIGC-supported digital learning platform, while the control group received traditional offline psychological nursing instruction. Key evaluation metrics included Mood Assessment Competence (MAC), Psychological Nursing Performance (PNP), and Learning Self-Efficacy Scale (LSES). Differences between groups were analyzed using a mixed-effects model, with thematic analysis identifying critical pathways for educational digitalization development. The experimental group demonstrated a 46% improvement in emotional assessment skills compared to baseline, significantly outperforming the control group's 19% increase (p=.002). Psychological nursing proficiency scores rose by 38% in the experimental group versus 14% in the control group, with statistically significant differences (p=.004). Learning self-efficacy improved by 41% in the experimental group, markedly higher than the control group's 17% increase (p=.006). Further interviews revealed three key benefits of AIGC: 1) Generating highly realistic emotional scenarios to enhance students’ grasp of emotional recognition essentials; 2) Personalizing learning paths through automated feedback; 3) Assisting instructors in developing scalable psychological nursing teaching resources. System requirements analysis identified four critical components for digitalized teaching pathways: content generation quality, ethical standards, bias prevention mechanisms, and teachers' digital literacy. Additionally, learners' platform usage frequency showed a positive correlation with skill improvement (p=.01), indicating that digital engagement significantly impacts training effectiveness. Research demonstrates that AI-assisted mental health education significantly enhances core competencies in emotional disorder care, particularly in critical skills such as emotional assessment, communication intervention, and crisis identification. Through immersive simulations, precise feedback, and resource scalability, AIGC provides a new technological pathway for digital mental health education. The development of educational digitalization should strike a balance between technological innovation and ethical safeguards, including content credibility, student privacy protection, and standardized application of emotional data. Overall, AIGC offers a viable approach to building intelligent, personalized, and sustainable educational systems for emotional disorder care, laying the foundation for future innovations in mental health education models. Future research could further integrate multimodal emotion recognition technology with generative AI systems to validate AIGC's long-term teaching effectiveness and adaptability in real-world clinical settings.

Virtual humans (i.e., embodied conversational agents) have the potential to support college students' mental health, particularly in Science, Technology, Engineering, and Mathematics (STEM) fields where students are at a heightened risk of mental disorders such as anxiety and depression. A comprehensive understanding of students, considering their cultural characteristics, experiences, and expectations, is crucial for creating timely and effective virtual human interventions. To this end, we conducted a user study with 481 computer science students from a major university in North America, exploring how they co-designed virtual humans to support mental health conversations for students similar to them. Our findings suggest that computer science students who engage in co-design processes of virtual humans tend to create agents that closely resemble them demographically--agent-designer demographic similarity. Key factors influencing virtual human design included age, gender, ethnicity, and the matching between appearance and voice. We also observed that the demographic characteristics of virtual human designers, especially ethnicity and gender, tend to be associated with those of the virtual humans they designed. Finally, we provide insights concerning the impact of user-designer demographic similarity in virtual humans' effectiveness in promoting mental health conversations when designers' characteristics are shared explicitly or implicitly. Understanding how virtual humans' characteristics serve users' experiences in mental wellness conversations and the similarity-attraction effects between agents, users, and designers may help tailor virtual humans' design to enhance their acceptance and increase their counseling effectiveness.

Medication literacy is integral to health literacy, pivotal for medication safety and adherence. It denotes an individual's capacity to discern, comprehend, and convey medication-related information. Existing scales, however, are time-consuming and predominantly cater to patients and community dwellers, necessitating a more succinct instrument. This study presents the development of a brief Medication Literacy Scale (MLS-14) utilizing classical test theory (CTT) and item response theory (IRT), targeting a college student demographic. The MLS-14's abbreviated version, a 6-item scale (MLS-SF), was distilled through CTT and IRT methodologies, engaging 2431 Chinese college students to scrutinize its psychometric properties. The MLS-SF demonstrated a Cronbach's α of 0.765, with three extracted factors via exploratory factor analysis, accounting for 66% of the cumulative variance. All items exhibited factor loadings above 0.5. The scale's three-factor structure was substantiated through confirmatory factor analysis with satisfactory fit indices (chi2/df=5.11, RMSEA=0.063, GFI=0.990, AGFI=0.966, NFI=0.984, IFI=0.987, CFI=0.987). IRT modeling confirmed reasonable discrimination and location parameters for all items, free of differential item functioning (DIF) by gender. Except for items 4 and 10, the remaining items were informative at medium theta levels, indicating their utility in assessing medication literacy efficiently. The developed 6-item Medication Literacy Short Form (MLS-SF) proves to be a reliable and valid instrument for the expedited evaluation of college students' medication literacy, offering a valuable addition to the arsenal of health literacy assessment tools.

Student mental health is a sensitive issue that necessitates special attention. A primary concern is the student-to-counselor ratio, which surpasses the recommended standard of 250:1 in most universities. This imbalance results in extended waiting periods for in-person consultations, which cause suboptimal treatment. Significant efforts have been directed toward developing mental health dialogue systems utilizing the existing open-source mental health-related datasets. However, currently available datasets either discuss general topics or various strategies that may not be viable for direct application due to numerous ethical constraints inherent in this research domain. To address this issue, this paper introduces a specialized mental health dataset that emphasizes the active listening strategy employed in conversation for counseling, also named as ConvCounsel. This dataset comprises both speech and text data, which can facilitate the development of a reliable pipeline for mental health dialogue systems. To demonstrate the utility of the proposed dataset, this paper also presents the NYCUKA, a spoken mental health dialogue system that is designed by using the ConvCounsel dataset. The results show the merit of using this dataset.

The rapid spread of health misinformation on online social networks (OSNs) during global crises such as the COVID-19 pandemic poses challenges to public health, social stability, and institutional trust. Centrality metrics have long been pivotal in understanding the dynamics of information flow, particularly in the context of health misinformation. However, the increasing complexity and dynamism of online networks, especially during crises, highlight the limitations of these traditional approaches. This study introduces and compares three novel centrality metrics: dynamic influence centrality (DIC), health misinformation vulnerability centrality (MVC), and propagation centrality (PC). These metrics incorporate temporal dynamics, susceptibility, and multilayered network interactions. Using the FibVID dataset, we compared traditional and novel metrics to identify influential nodes, propagation pathways, and misinformation influencers. Traditional metrics identified 29 influential nodes, while the new metrics uncovered 24 unique nodes, resulting in 42 combined nodes, an increase of 44.83%. Baseline interventions reduced health misinformation by 50%, while incorporating the new metrics increased this to 62.5%, an improvement of 25%. To evaluate the broader applicability of the proposed metrics, we validated our framework on a second dataset, Monant Medical Misinformation, which covers a diverse range of health misinformation discussions beyond COVID-19. The results confirmed that the advanced metrics generalised successfully, identifying distinct influential actors not captured by traditional methods. In general, the findings suggest that a combination of traditional and novel centrality measures offers a more robust and generalisable framework for understanding and mitigating the spread of health misinformation in different online network contexts.

The COVID-19 pandemic has highlighted the dire necessity to improve public health literacy for societal resilience. YouTube, the largest video-sharing social media platform, provides a vast repository of user-generated health information in a multi-media-rich format which may be easier for the public to understand and use if major concerns about content quality and accuracy are addressed. This study develops an automated solution to identify, retrieve and shortlist medically relevant and understandable YouTube videos that domain experts can subsequently review and recommend for disseminating and educating the public on the COVID-19 pandemic and similar public health outbreaks. Our approach leverages domain knowledge from human experts and machine learning and natural language processing methods to provide a scalable, replicable, and generalizable approach that can also be applied to enhance the management of many health conditions.

In this paper, we present a comprehensive survey on the pervasive issue of medical misinformation in social networks from the perspective of information technology. The survey aims at providing a systematic review of related research and helping researchers and practitioners navigate through this fast-changing field. Specifically, we first present manual and automatic approaches for fact-checking. We then explore fake news detection methods, using content, propagation features, or source features, as well as mitigation approaches for countering the spread of misinformation. We also provide a detailed list of several datasets on health misinformation and of publicly available tools. We conclude the survey with a discussion on the open challenges and future research directions in the battle against health misinformation.

The chapter discusses the foundational impact of modern generative AI models on information access (IA) systems. In contrast to traditional AI, the large-scale training and superior data modeling of generative AI models enable them to produce high-quality, human-like responses, which brings brand new opportunities for the development of IA paradigms. In this chapter, we identify and introduce two of them in details, i.e., information generation and information synthesis. Information generation allows AI to create tailored content addressing user needs directly, enhancing user experience with immediate, relevant outputs. Information synthesis leverages the ability of generative AI to integrate and reorganize existing information, providing grounded responses and mitigating issues like model hallucination, which is particularly valuable in scenarios requiring precision and external knowledge. This chapter delves into the foundational aspects of generative models, including architecture, scaling, and training, and discusses their applications in multi-modal scenarios. Additionally, it examines the retrieval-augmented generation paradigm and other methods for corpus modeling and understanding, demonstrating how generative AI can enhance information access systems. It also summarizes potential challenges and fruitful directions for future studies.

Generative AI (GenAI) models have demonstrated remarkable capabilities in a wide variety of medical tasks. However, as these models are trained using generalist datasets with very limited human oversight, they can learn uses of medical products that have not been adequately evaluated for safety and efficacy, nor approved by regulatory agencies. Given the scale at which GenAI may reach users, unvetted recommendations pose a public health risk. In this work, we propose an approach to identify potentially harmful product recommendations, and demonstrate it using a recent multimodal large language model.

Artificial intelligence (AI) models trained using medical images for clinical tasks often exhibit bias in the form of disparities in performance between subgroups. Since not all sources of biases in real-world medical imaging data are easily identifiable, it is challenging to comprehensively assess how those biases are encoded in models, and how capable bias mitigation methods are at ameliorating performance disparities. In this article, we introduce a novel analysis framework for systematically and objectively investigating the impact of biases in medical images on AI models. We developed and tested this framework for conducting controlled in silico trials to assess bias in medical imaging AI using a tool for generating synthetic magnetic resonance images with known disease effects and sources of bias. The feasibility is showcased by using three counterfactual bias scenarios to measure the impact of simulated bias effects on a convolutional neural network (CNN) classifier and the efficacy of three bias mitigation strategies. The analysis revealed that the simulated biases resulted in expected subgroup performance disparities when the CNN was trained on the synthetic datasets. Moreover, reweighing was identified as the most successful bias mitigation strategy for this setup, and we demonstrated how explainable AI methods can aid in investigating the manifestation of bias in the model using this framework. Developing fair AI models is a considerable challenge given that many and often unknown sources of biases can be present in medical imaging datasets. In this work, we present a novel methodology to objectively study the impact of biases and mitigation strategies on deep learning pipelines, which can support the development of clinical AI that is robust and responsible.

While Large Language Models (LLMs) can generate fluent and convincing responses, they are not necessarily correct. This is especially apparent in the popular decompose-then-verify factuality evaluation pipeline, where LLMs evaluate generations by decomposing the generations into individual, valid claims. Factuality evaluation is especially important for medical answers, since incorrect medical information could seriously harm the patient. However, existing factuality systems are a poor match for the medical domain, as they are typically only evaluated on objective, entity-centric, formulaic texts such as biographies and historical topics. This differs from condition-dependent, conversational, hypothetical, sentence-structure diverse, and subjective medical answers, which makes decomposition into valid facts challenging. We propose MedScore, a new pipeline to decompose medical answers into condition-aware valid facts and verify against in-domain corpora. Our method extracts up to three times more valid facts than existing methods, reducing hallucination and vague references, and retaining condition-dependency in facts. The resulting factuality score substantially varies by decomposition method, verification corpus, and used backbone LLM, highlighting the importance of customizing each step for reliable factuality evaluation by using our generalizable and modularized pipeline for domain adaptation.

This paper introduces Med-Bot, an AI-powered chatbot designed to provide users with accurate and reliable medical information. Utilizing advanced libraries and frameworks such as PyTorch, Chromadb, Langchain and Autogptq, Med-Bot is built to handle the complexities of natural language understanding in a healthcare context. The integration of llamaassisted data processing and AutoGPT-Q provides enhanced performance in processing and responding to queries based on PDFs of medical literature, ensuring that users receive precise and trustworthy information. This research details the methodologies employed in developing Med-Bot and evaluates its effectiveness in disseminating healthcare information.

While generative artificial intelligence (Gen AI) increasingly transforms academic environments, a critical gap exists in understanding and mitigating human biases in AI interactions, such as anchoring and confirmation bias. This position paper advocates for metacognitive AI literacy interventions to help university students critically engage with AI and address biases across the Human-AI interaction workflows. The paper presents the importance of considering (1) metacognitive support with deliberate friction focusing on human bias; (2) bi-directional Human-AI interaction intervention addressing both input formulation and output interpretation; and (3) adaptive scaffolding that responds to diverse user engagement patterns. These frameworks are illustrated through ongoing work on "DeBiasMe," AIED (AI in Education) interventions designed to enhance awareness of cognitive biases while empowering user agency in AI interactions. The paper invites multiple stakeholders to engage in discussions on design and evaluation methods for scaffolding mechanisms, bias visualization, and analysis frameworks. This position contributes to the emerging field of AI-augmented learning by emphasizing the critical role of metacognition in helping students navigate the complex interaction between human, statistical, and systemic biases in AI use while highlighting how cognitive adaptation to AI systems must be explicitly integrated into comprehensive AI literacy frameworks.

Trust is often cited as an essential criterion for the effective use and real-world deployment of AI. Researchers argue that AI should be more transparent to increase trust, making transparency one of the main goals of XAI. Nevertheless, empirical research on this topic is inconclusive regarding the effect of transparency on trust. An explanation for this ambiguity could be that trust is operationalized differently within XAI. In this position paper, we advocate for a clear distinction between behavioral (objective) measures of reliance and attitudinal (subjective) measures of trust. However, researchers sometimes appear to use behavioral measures when intending to capture trust, although attitudinal measures would be more appropriate. Based on past research, we emphasize that there are sound theoretical reasons to keep trust and reliance separate. Properly distinguishing these two concepts provides a more comprehensive understanding of how transparency affects trust and reliance, benefiting future XAI research.

As AI-generated health information proliferates online and becomes increasingly indistinguishable from human-sourced information, it becomes critical to understand how people trust and label such content, especially when the information is inaccurate. We conducted two complementary studies: (1) a mixed-methods survey (N=142) employing a 2 (source: Human vs. LLM) $\times$ 2 (label: Human vs. AI) $\times$ 3 (type: General, Symptom, Treatment) design, and (2) a within-subjects lab study (N=40) incorporating eye-tracking and physiological sensing (ECG, EDA, skin temperature). Participants were presented with health information varying by source-label combinations and asked to rate their trust, while their gaze behavior and physiological signals were recorded. We found that LLM-generated information was trusted more than human-generated content, whereas information labeled as human was trusted more than that labeled as AI. Trust remained consistent across information types. Eye-tracking and physiological responses varied significantly by source and label. Machine learning models trained on these behavioral and physiological features predicted binary self-reported trust levels with 73% accuracy and information source with 65% accuracy. Our findings demonstrate that adding transparency labels to online health information modulates trust. Behavioral and physiological features show potential to verify trust perceptions and indicate if additional transparency is needed.

Building trust in AI-based systems is deemed critical for their adoption and appropriate use. Recent research has thus attempted to evaluate how various attributes of these systems affect user trust. However, limitations regarding the definition and measurement of trust in AI have hampered progress in the field, leading to results that are inconsistent or difficult to compare. In this work, we provide an overview of the main limitations in defining and measuring trust in AI. We focus on the attempt of giving trust in AI a numerical value and its utility in informing the design of real-world human-AI interactions. Taking a socio-technical system perspective on AI, we explore two distinct approaches to tackle these challenges. We provide actionable recommendations on how these approaches can be implemented in practice and inform the design of human-AI interactions. We thereby aim to provide a starting point for researchers and designers to re-evaluate the current focus on trust in AI, improving the alignment between what empirical research paradigms may offer and the expectations of real-world human-AI interactions.

Artificial Intelligence-Generated Content (AIGC) is a rapidly evolving field that utilizes advanced AI algorithms to generate content. Through integration with mobile edge networks, mobile AIGC networks have gained significant attention, which can provide real-time customized and personalized AIGC services and products. Since blockchains can facilitate decentralized and transparent data management, AIGC products can be securely managed by blockchain to avoid tampering and plagiarization. However, the evolution of blockchain-empowered mobile AIGC is still in its nascent phase, grappling with challenges such as improving information propagation efficiency to enable blockchain-empowered mobile AIGC. In this paper, we design a Graph Attention Network (GAT)-based information propagation optimization framework for blockchain-empowered mobile AIGC. We first innovatively apply age of information as a data-freshness metric to measure information propagation efficiency in public blockchains. Considering that GATs possess the excellent ability to process graph-structured data, we utilize the GAT to obtain the optimal information propagation trajectory. Numerical results demonstrate that the proposed scheme exhibits the most outstanding information propagation efficiency compared with traditional routing mechanisms.

After the release of several widely adopted artificial intelligence (AI) literacy guidelines by 2021, the unprecedented rise of generative AI since 2023 has transformed the way we work and acquire information worldwide. Unlike traditional AI algorithms, generative AI exhibits distinct and more nuanced characteristics. However, a lack of robust understanding of generative AI hinders individuals' ability to use generative AI effectively, critically, and responsibly, which we can call generative AI literacy. To address this gap, we reviewed and synthesized existing literature and proposed generative AI literacy guidelines with 12 items organized into four aspects: (1) generative AI tool selection and prompting, (2) understanding interaction with generative AI, (3) understanding generative AI outputs, and (4) high-level understanding of generative AI technologies. These guidelines aim to support schools, companies, and organizations in developing frameworks that support their members to use generative AI in an efficient, ethical, and informed way.

Deepfakes images can erode trust in institutions and compromise election outcomes, as people often struggle to discern real images from deepfake images. Improving digital literacy can help address these challenges. Here, we compare the efficacy of five digital literacy interventions to boost people's ability to discern deepfakes: (1) textual guidance on common indicators of deepfakes; (2) visual demonstrations of these indicators; (3) a gamified exercise for identifying deepfakes; (4) implicit learning through repeated exposure and feedback; and (5) explanations of how deepfakes are generated with the help of AI. We conducted an experiment with N=1,200 participants from the United States to test the immediate and long-term effectiveness of our interventions. Our results show that our lightweight, easy-to-understand interventions can boost deepfake image discernment by up to 13 percentage points while maintaining trust in real images.

Even though AI literacy has emerged as a prominent education topic in the wake of generative AI, its definition remains vague. There is little consensus among researchers and practitioners on how to discuss and design AI literacy interventions. The term has been used to describe both learning activities that train undergraduate students to use ChatGPT effectively and having kindergarten children interact with social robots. This paper applies an integrative review method to examine empirical and theoretical AI literacy studies published since 2020. In synthesizing the 124 reviewed studies, three ways to conceptualize literacy-functional, critical, and indirectly beneficial-and three perspectives on AI-technical detail, tool, and sociocultural-were identified, forming a framework that reflects the spectrum of how AI literacy is approached in practice. The framework highlights the need for more specialized terms within AI literacy discourse and indicates research gaps in certain AI literacy objectives.

This paper introduces a competency-based model for generative artificial intelligence (AI) literacy covering essential skills and knowledge areas necessary to interact with generative AI. The competencies range from foundational AI literacy to prompt engineering and programming skills, including ethical and legal considerations. These twelve competencies offer a framework for individuals, policymakers, government officials, and educators looking to navigate and take advantage of the potential of generative AI responsibly. Embedding these competencies into educational programs and professional training initiatives can equip individuals to become responsible and informed users and creators of generative AI. The competencies follow a logical progression and serve as a roadmap for individuals seeking to get familiar with generative AI and for researchers and policymakers to develop assessments, educational programs, guidelines, and regulations.

The cancer knowledge gap represents a significant disparity in awareness and understanding of cancer-related information across different demographic groups. Leveraging Artificial Intelligence-Generated Content (AIGC) offers a promising approach to personalize health education and potentially bridge this gap. This study aimed to evaluate the potential of AIGC to bridge the cancer knowledge gap, assessing the roles of health self-efficacy as a mediator and educational level as a moderator in this relationship. A 6-month longitudinal study was conducted using online surveys distributed to undergraduate students in non-medical disciplines at one university and graduate students in medical specialties at another university in China. The study assessed the frequency of AIGC use, health self-efficacy, and cancer knowledge at two time points. The results indicated that AIGC use significantly enhanced cancer knowledge levels and health self-efficacy over time. Educational level notably moderated the effects of AIGC use, with non-medical undergraduate students showing greater gains in knowledge and self-efficacy. Additionally, health self-efficacy mediated the relationship between AIGC use and cancer knowledge, underscoring the importance of health self-efficacy. The study confirms the efficacy of AIGC in narrowing the cancer knowledge gap and enhancing health self-efficacy, particularly among students with lower initial medical knowledge. These findings highlight the potential of integrating AIGC tools in cancer education and public health interventions, particularly for individuals at different educational levels. By tailoring digital health resources to varying educational needs, these interventions could enhance cancer knowledge acquisition, improve health self-efficacy, and contribute to better cancer prevention and control outcomes. This study explores whether artificial intelligence-generated content (AIGC) tools, such as ChatGPT, can reduce disparities in cancer knowledge across different education levels. The research focused on university students in China over six months, with one group studying computer science and another in medical fields. Participants used AI tools to learn about cancer prevention, early detection, and treatment. The study found that frequent use of AIGC tools significantly improved participants' understanding of cancer and their confidence in managing their health (health self-efficacy). Non-medical students benefited the most, as they started with less prior knowledge compared to medical students. Additionally, students who felt more confident in their health knowledge were more likely to understand and retain cancer-related information. This research shows that AIGC tools can bridge knowledge gaps by providing personalized and accessible content. The findings suggest that such tools can play a vital role in public health campaigns and educational programs, especially for people with limited access to traditional healthcare education. These insights highlight the potential for AI technology to improve cancer awareness and health decision-making, ultimately contributing to better prevention and treatment outcomes.

The aim of this study was to investigate the perceptions of health profession students regarding ChatGPT use and the potential impact of integrating ChatGPT in healthcare and education. Artificial Intelligence is increasingly utilized in medical education and clinical profession training. However, since its introduction, ChatGPT remains relatively unexplored in terms of health profession students' acceptance of its use in education and practice. This study employed a mixed-methods approach, using a web-based survey. The study involved a convenience sample recruited through various methods, including Faculty of Medicine announcements, social media, and snowball sampling, during the second semester (March to June 2023). Data were collected using a structured questionnaire with closed-ended questions and three open-ended questions. The final sample comprised 217 undergraduate health profession students, including 73 (33.6%) nursing students, 65 (30.0%) medical students, and 79 (36.4%) occupational therapy, physiotherapy, and speech therapy students. Among the surveyed students, 86.2% were familiar with ChatGPT, with generally positive perceptions as reflected by a mean score of 4.04 (SD = 0.62) on a scale of 1 to 5. Positive feedback was particularly noted with respect to ChatGPT's role in information retrieval and summarization. The qualitative data revealed three main themes: experiences with ChatGPT, its impact on the quality of healthcare, and its integration into the curriculum. The findings highlight benefits such as serving as a convenient tool for accessing information, reducing human errors, and fostering innovative learning approaches. However, they also underscore areas of concern, including ethical considerations, challenges in fostering critical thinking, and issues related to verification. The absence of significant differences between the different fields of study indicates consistent perceptions across nursing, medicine, and other health profession students. Our findings underscore the necessity for continuous refinement to enhance ChatGPT's accuracy, reliability, and alignment with the diverse educational needs of health professions. These insights not only deepen our understanding of student perceptions of ChatGPT in healthcare education but also have significant implications for the future integration of AI in health profession practice. The study emphasizes the importance of a careful balance between leveraging the benefits of AI tools and addressing ethical and pedagogical concerns.

The integration of generative artificial intelligence (AI) tools such as ChatGPT into healthcare education has increased significantly in recent years. These tools are frequently used by students to access medical knowledge, practice clinical reasoning, and supplement coursework. However, concerns remain regarding the accuracy, reliability, and educational value of AI-generated health information. Existing literature highlights both the potential and the limitations of these tools, yet limited empirical evidence is available concerning students' perceptions and trust in such systems, particularly within the Saudi Arabian context. This study aimed to evaluate the trustworthiness of ChatGPT-generated health information from the perspective of future healthcare professionals in Saudi Arabia, and to identify the factors influencing their willingness to adopt such tools in academic and clinical settings. A cross-sectional survey design was employed, targeting undergraduate students enrolled in health sciences programs across selected Saudi universities. A structured, self-administered questionnaire was used to measure demographic variables, knowledge of generative AI, and attitudes based on the Technology Acceptance Model. A total of 518 responses were analyzed using descriptive statistics, Pearson's correlation, and multiple linear regression. Participants demonstrated moderate trust in ChatGPT for health-related queries (M=3.15, SD=0.78), with high perceived importance placed on expert verification and source citation. In the multiple linear regression analysis (N=284), perceived reliability (B=0.42, p<.001) and perceived accuracy (B=0.26, p<.001) emerged as the strongest positive predictors of willingness to recommend the tool, whereas risk awareness had a significant negative association (B=-0.19, p=.002). The findings of this study indicate that undergraduate health sciences students in Saudi Arabia hold a cautiously optimistic view of ChatGPT as a supplementary tool for health-related learning. While many participants recognized its usefulness, their willingness to rely on or recommend the tool was closely linked to how reliable and accurate they perceived its content to be. The emphasis placed on expert validation and credible sources underscores a broader need to integrate digital literacy and critical appraisal skills into health education curricula, particularly as AI becomes more embedded in academic practice.

BackgroundChatbots are increasingly integrated into healthcare, offering personalized and accessible health advice. However, the impact of factors such as chatbot authority, health information type, and interaction style on users' decision-making remains unclear.ObjectiveThis study aims to investigate how these elements influence users' willingness to adopt health advice provided by chatbots.MethodsA 2 × 2 × 2 factorial experiment was conducted with 480 university students to examine the effects of chatbot authority (authoritative vs. non-authoritative), health information type (preventive vs. treatment-related), and interaction style (formal vs. informal). Participants' willingness to adopt the health advice was measured before and after interacting with the chatbot.ResultsThe study found that a authoritative chatbot delivering treatment-related advice in a formal style significantly increased willingness to adopt the advice. Conversely, preventive information was more effective when presented informally by a non-authoritative chatbot. These results support the media evocation paradigm, which suggests that chatbots framed as authoritative figures evoke greater user engagement and trust in health contexts.ConclusionThe findings extend the media evocation paradigm by demonstrating that chatbot authority, information type, and interaction style should be aligned with the nature of health advice to maximize effectiveness. This study provides insights for designing chatbots that improve health decision-making by tailoring their communication strategies.

AI-Generated Content (AIGC) is reshaping information landscapes in Africa and Latin America, impacting women's Sexual and Reproductive Health and Rights (SRHR) cognition. This paper explores AIGC's dualistic role through systematic literature synthesis, critical analysis of reports from international health organizations and government agencies, and the development of illustrative case scenarios grounded in documented experiences from these regions. Rather than presenting primary ethnographic data, this conceptual analysis synthesizes existing evidence and constructs contextually informed vignettes to illuminate the complex interplay between emerging technology, cultural contexts, and health cognition. The study highlights the digital divide, cultural factors, and trust as crucial mediators, demonstrating AIGC's dual potential: a tool for empowerment by improving information accessibility, personalization, and discussion of sensitive issues, yet also a risk for exacerbating inequalities through misinformation, algorithmic bias, and ethical concerns. It underscores women's active negotiation of AIGC within their specific socio-cultural contexts. The paper proposes strategic recommendations for technology developers, health systems, and policymakers to responsibly leverage AIGC, promoting a human-centric, equitable, and culturally sensitive approach to foster health equity. Le contenu généré par l'IA (CGIA) remodèle le paysage informationnel en Afrique et en Amérique latine, influençant la perception des femmes en matière de santé et de droits sexuels et reproductifs (SDSR). Cet article explore le double rôle du CGIA à travers une synthèse systématique de la littérature, une analyse critique des rapports d'organisations internationales de santé et d'agences gouvernementales, et l'élaboration de scénarios illustratifs fondés sur des expériences documentées dans ces régions. Plutôt que de présenter des données ethnographiques primaires, cette analyse conceptuelle synthétise les données existantes et construit des vignettes contextualisées pour éclairer l'interaction complexe entre les technologies émergentes, les contextes culturels et la perception de la santé. L'étude met en lumière la fracture numérique, les facteurs culturels et la confiance comme médiateurs essentiels, démontrant le double potentiel du CGIA : un outil d'émancipation grâce à une meilleure accessibilité à l'information, une personnalisation accrue et une discussion plus approfondie des sujets sensibles ; mais aussi un risque d'exacerbation des inégalités par la désinformation, les biais algorithmiques et les problèmes éthiques. Elle souligne la manière dont les femmes s'approprient activement le CGIA dans leurs contextes socioculturels spécifiques. Ce document propose des recommandations stratégiques à l'intention des développeurs de technologies, des systèmes de santé et des décideurs politiques afin d'exploiter de manière responsable l'IAGC, en promouvant une approche centrée sur l'humain, équitable et respectueuse des différences culturelles pour favoriser l'équité en santé.

Access to patient-centered health information is essential for informed decision-making. However, online medical resources vary in quality and often fail to accommodate differing degrees of health literacy. This issue is particularly evident in surgical contexts, where complex terminology obstructs patient comprehension. With the increasing reliance on AI models for supplementary medical information, the reliability and readability of AI-generated content require thorough evaluation. This study aimed to evaluate four natural language processing models-ChatGPT-4o, ChatGPT-o3 mini, DeepSeek-V3, and DeepSeek-R1-in generating patient education materials for three common spinal surgeries: lumbar discectomy, spinal fusion, and decompressive laminectomy. Information quality was evaluated using the DISCERN score, and readability was assessed through Flesch-Kincaid indices. DeepSeek-R1 produced the most readable responses, with Flesch-Kincaid Grade Level (FKGL) scores ranging from 7.2 to 9.0, succeeded by ChatGPT-4o. In contrast, ChatGPT-o3 exhibited the lowest readability (FKGL > 10.4). The DISCERN scores for all AI models were below 60, classifying the information quality as "fair," primarily due to insufficient cited references. All models achieved merely a "fair" quality rating, underscoring the necessity for improvements in citation practices, and personalization. Nonetheless, DeepSeek-R1 and ChatGPT-4o generated more readable surgical information than ChatGPT-o3. Given that enhanced readability can improve patient engagement, reduce anxiety, and contribute to better surgical outcomes, these two models should be prioritized for assisting patients in the clinical. This study is limited by the rapid evolution of AI models, its exclusive focus on spinal surgery education, and the absence of real-world patient feedback, which may affect the generalizability and long-term applicability of the findings. Future research ought to explore interactive, multimodal approaches and incorporate patient feedback to ensure that AI-generated health information is accurate, accessible, and facilitates informed healthcare decisions.

Conversational artificial intelligence, in the form of chatbots powered by large language models, offers a new approach to facilitating human-like interactions, yet its efficacy in enhancing vaccination uptake remains under-investigated. This study assesses the effectiveness of a vaccine chatbot in improving human papillomavirus (HPV) vaccination among female middle school students aged 12-15 years across diverse socioeconomic settings in China, where HPV vaccination is primarily paid out-of-pocket. A school-based cluster randomized trial was conducted from 18 January to 31 May 2024. The study included 2,671 parents from 180 middle school classes stratified by socioeconomic setting, school and grade level in Shanghai megacity, and urban and rural regions of Anhui Province. Participants were randomly assigned to either the intervention group (90 classes, 1,294 parents), which engaged with the chatbot for two weeks, or the control group (90 classes, 1,377 parents), which received usual care. The primary outcome was the receipt or scheduled appointment of the HPV vaccine for participants' daughters. In intention-to-treat analyses, 7.1% of the intervention group met this outcome versus 1.8% of the control group (P < 0.001) over a two-week intervention period. In addition, there was a statistically significant increase in HPV vaccination-specific consultations with health professionals (49.1% versus 17.6%, P < 0.001), along with enhanced vaccine literacy (P < 0.001) and rumor discernment (P < 0.001) among participants using the chatbot. These findings indicate that the chatbot effectively increased vaccination and improved parental vaccine literacy, although further research is necessary to scale and sustain these gains. Clinical trial registration: NCT06227689 .

Artificial intelligence (AI) chatbots are increasingly used for medical inquiries, including sensitive topics like sexually transmitted diseases (STDs). However, concerns remain regarding the reliability and readability of the information they provide. This study aimed to assess the reliability and readability of AI chatbots in providing information on STDs. The key objectives were to determine (1) the reliability of STD-related information provided by AI chatbots, and (2) whether the readability of this information meets the recommended standarts for patient education materials. Eleven relevant STD-related search queries were identified using Google Trends and entered into four AI chatbots: ChatGPT, Gemini, Perplexity, and Copilot. The reliability of the responses was evaluated using established tools, including DISCERN, EQIP, JAMA, and GQS. Readability was assessed using six widely recognized metrics, such as the Flesch-Kincaid Grade Level and the Gunning Fog Index. The performance of chatbots was statistically compared in terms of reliability and readability. The analysis revealed significant differences in reliability across the AI chatbots. Perplexity and Copilot consistently outperformed ChatGPT and Gemini in DISCERN and EQIP scores, suggesting that these two chatbots provided more reliable information. However, results showed that none of the chatbots achieved the 6th-grade readability standard. All the chatbots generated information that was too complex for the general public, especially for individuals with lower health literacy levels. While Perplexity and Copilot showed better reliability in providing STD-related information, none of the chatbots met the recommended readability benchmarks. These findings highlight the need for future improvements in both the accuracy and accessibility of AI-generated health information, ensuring it can be easily understood by a broader audience.

The integration of generative AI (GenAI) into academic workflows represents a fundamental shift in scientific practice. While these tools can amplify productivity, they risk eroding the cognitive foundations of expertise by simulating the very tasks through which scientific competence is developed, from synthesis to experimental design to writing. Uncritical reliance can lead to skill atrophy and AI complacency. We propose a framework of essential AI meta-skills: strategic direction, critical discernment, and systematic calibration. These constitute a new form of scientific literacy that builds on traditional critical thinking. Through domain-specific examples and a pedagogical model based on situated learning, we show how these meta-skills can be cultivated to ensure that researchers, particularly trainees, maintain intellectual autonomy. Without deliberate cultivation of these meta-skills, we risk creating the first generation of researchers who serve their tools rather than direct them.

Generative artificial intelligence (AI) chatbots have recently been posited as potential sources of online medical information for patients making medical decisions. Existing online patient-oriented medical information has repeatedly been shown to be of variable quality and difficult readability. Therefore, we sought to evaluate the content and quality of AI-generated medical information on acute appendicitis. A modified DISCERN assessment tool, comprising 16 distinct criteria each scored on a 5-point Likert scale (score range 16-80), was used to assess AI-generated content. Readability was determined using the Flesch Reading Ease (FRE) and Flesch-Kincaid Grade Level (FKGL) scores. Four popular chatbots, ChatGPT-3.5 and ChatGPT-4, Bard, and Claude-2, were prompted to generate medical information about appendicitis. Three investigators independently scored the generated texts blinded to the identity of the AI platforms. ChatGPT-3.5, ChatGPT-4, Bard, and Claude-2 had overall mean (SD) quality scores of 60.7 (1.2), 62.0 (1.0), 62.3 (1.2), and 51.3 (2.3), respectively, on a scale of 16-80. Inter-rater reliability was 0.81, 0.75, 0.81, and 0.72, respectively, indicating substantial agreement. Claude-2 demonstrated a significantly lower mean quality score compared to ChatGPT-4 (p = 0.001), ChatGPT-3.5 (p = 0.005), and Bard (p = 0.001). Bard was the only AI platform that listed verifiable sources, while Claude-2 provided fabricated sources. All chatbots except for Claude-2 advised readers to consult a physician if experiencing symptoms. Regarding readability, FKGL and FRE scores of ChatGPT-3.5, ChatGPT-4, Bard, and Claude-2 were 14.6 and 23.8, 11.9 and 33.9, 8.6 and 52.8, 11.0 and 36.6, respectively, indicating difficulty readability at a college reading skill level. AI-generated medical information on appendicitis scored favorably upon quality assessment, but most either fabricated sources or did not provide any altogether. Additionally, overall readability far exceeded recommended levels for the public. Generative AI platforms demonstrate measured potential for patient education and engagement about appendicitis.

Artificial intelligence (AI) and the introduction of Large Language Model (LLM) chatbots have become a common source of patient inquiry in healthcare. The quality and readability of AI-generated patient education materials (PEM) is the subject of many studies across multiple medical topics. Most demonstrate poor readability and acceptable quality. However, an area yet to be investigated is chemotherapy-induced cardiotoxicity. This study seeks to assess the quality and readability of chatbot created PEM relative to chemotherapy-induced cardiotoxicity. We conducted an observational cross-sectional study in August 2024 by asking 10 questions to 4 chatbots: ChatGPT, Microsoft Copilot (Copilot), Google Gemini (Gemini), and Meta AI (Meta). The generated material was assessed for readability using 7 tools: Flesch Reading Ease Score (FRES), Flesch-Kincaid Grade Level (FKGL), Gunning Fog Index (GFI), Coleman-Liau Index (CLI), Simple Measure of Gobbledygook (SMOG) Index, Automated Readability Index (ARI), and FORCAST Grade Level. Quality was assessed using modified versions of 2 validated tools: the Patient Education Materials Assessment Tool (PEMAT), which outputs a 0% to 100% score, and DISCERN, a 1 (unsatisfactory) to 5 (highly satisfactory) scoring system. Descriptive statistics were used to evaluate performance and compare chatbots amongst each other. Mean reading grade level (RGL) across all chatbots was 13.7. Calculated RGLs for ChatGPT, Copilot, Gemini and Meta were 14.2, 14.0, 12.5, 14.2, respectively. Mean DISCERN scores across the chatbots was 4.2. DISCERN scores for ChatGPT, Copilot, Gemini, and Meta were 4.2, 4.3, 4.2, and 3.9, respectively. Median PEMAT scores for understandability and actionability were 91.7% and 75%, respectively. Understandability and actionability scores for ChatGPT, Copilot, Gemini, and Meta were 100% and 75%, 91.7% and 75%, 90.9% and 75%, and 91.7% and 50%, respectively. AI chatbots produce high quality PEM with poor readability. We do not discourage using chatbots to create PEM but recommend cautioning patients about their readability concerns. AI chatbots are not an alternative to a healthcare provider. Furthermore, there is no consensus on which chatbots create the highest quality PEM. Future studies are needed to assess the effectiveness of AI chatbots in providing PEM to patients and how the capabilities of AI chatbots are changing over time.

ChatGPT has gained significant popularity as a source of healthcare information among the general population. Evaluating the quality of chatbot responses is crucial, requiring comprehensive and qualitative analysis. This study aims to assess the answers provided by ChatGPT during hypothetical breast augmentation consultations across various categories and depths. The evaluation involves the utilization of validated tools and a comparison of scores between plastic surgeons and laypersons. A panel consisting of five plastic surgeons and five laypersons evaluated ChatGPT's responses to 25 questions spanning consultation, procedure, recovery, and sentiment categories. The DISCERN and PEMAT tools were employed to assess the responses, while emotional context was examined through ten specific questions. Additionally, readability was measured using the Flesch Reading Ease score. Qualitative analysis was performed to identify the overall strengths and weaknesses. Plastic surgeons generally scored lower than laypersons across most domains. Scores for each evaluation domain varied by category, with the consultation category demonstrating lower scores in terms of DISCERN reliability, information quality, and DISCERN score. Plastic surgeons assigned significantly lower overall quality ratings to the procedure category compared to other question categories. They also gave lower emotion scores in the procedure category compared to laypersons. The depth of the questions did not impact the scoring. Existing health information evaluation tools may not be entirely suitable for comprehensively evaluating the quality of individual responses generated by ChatGPT. Consequently, the development and implementation of appropriate evaluation tools to assess the appropriateness and quality of AI consultations are necessary.

Large language models (LLMs), a core technology of generative artificial intelligence (AI), are increasingly used in health education and promotion. Although they may expand access to medical information, concerns remain regarding the reliability and readability of AI generated content for the public. This study evaluated the reliability and readability of answers generated by five LLMs to common questions about perinatal depression. The primary aims were to determine (1) the reliability of LLM responses to frequently asked questions about perinatal depression and (2) whether the readability of the generated content aligns with public health literacy levels. Twenty-seven frequently asked questions were derived from Google Trends and patient facing resources from the American College of Obstetricians and Gynecologists (ACOG). Each question was submitted to ChatGPT-5, Gemini-2.5, Microsoft Copilot, Grok4, and DeepSeek. Two obstetricians independently rated responses using five validated instruments (DISCERN, EQIP, JAMA, GQS, and HONCODE) and inter-rater agreement was quantified using the interclass correlation coefficient (ICC). Readability was assessed using six indices: ARI, GFI, CLI, OLWF, LWGLF, and FRF. Differences among models were analyzed using the Friedman test. Inter rater agreement was high across 27 perinatal depression questions. ICC values ranged from 0.729 to 0.847. Significant between model differences emerged for DISCERN, EQIP, and HONCODE. All had Most LLMs demonstrated moderate to high reliability when responding to perinatal depression questions, supporting their potential as supplementary sources of health information. However, readability levels above recommended benchmarks suggest that current outputs may remain challenging for individuals with lower health literacy. While LLMs improve information accessibility, further improvements in readability, source attribution, and ethical transparency are needed to maximize public benefit and support equitable health communication. Future work should focus on defining and standardizing safety behaviors in high-risk mental health contexts to enable reliable clinical deployment.

No abstract

This study assessed the accuracy, quality, and readability of responses from three leading AI chatbots-ChatGPT-3.5, DeepSeek-V3, and Google Gemini-2.5-on the diagnosis, treatment, and long-term risks of adult hypothyroidism, comparing their outputs with current clinical guidelines. Two thyroid specialists developed 27 questions based on the Guideline for the Diagnosis and Management of Hypothyroidism in Adults (2017 edition), covering three categories: diagnosis, treatment, and long-term health risks. Responses from each AI model were independently evaluated by two reviewers. Accuracy was rated using a six-point Likert scale, quality using the DISCERN tool and the five-point Likert scale, and readability was assessed by the Flesch Reading Ease (FRE), Flesch-Kincaid Grade Level (FKGL), Gunning Fog Index (GFI),and Simple Measure of Gobbledygook(SMOG). All three AI models demonstrated excellent performance in accuracy (mean score > 4.5) and quality (high-quality rate > 94%). According to the DISCERN tool, no significant difference was observed in the overall information quality among the models. However, Gemini-2.5 generated responses of significantly lower quality for treatment-related questions than for diagnostic inquiries. The content generated by all models was relatively difficult to comprehend (low FRE scores and high FKGL/GFI scores), generally requiring a college-level or higher education for adequate understanding. All three AI chatbots were capable of producing highly accurate and high-quality medical information regarding hypothyroidism, with their responses showing strong consistency with clinical guidelines. This underscores the substantial potential of AI in supporting medical information delivery. However, the consistently high reading difficulty of their outputs may limit their practical utility in patient education. Future research should focus on improving the readability and patient-friendliness of AI outputs-through prompt engineering and multi-round dialogue optimization-while maintaining professional accuracy, to enable broader application of AI in health education.

This study aims to evaluate the accuracy and reliability of Generative Pre-trained Transformer (ChatGPT; OpenAI, San Francisco, California) in answering patient-related questions about trigger finger. This evaluation has the potential to enhance patient education prior to treatment and provides insight into the role of artificial intelligence (AI)-based systems in the patient educa-tion process. The ten most frequently asked questions regarding trigger finger were compiled from patient education websites and a literature review, then posed to ChatGPT. Two orthopedic specialists evaluated the responses using the Journal of the American Medical Association (JAMA) Benchmark criteria and the DISCERN instrument (A Tool for Judging the Quality of Written Consumer Health Information on Treatment Choices). Additionally, the readability of the responses was assessed using the Flesch-Kincaid Grade Level. The DISCERN scores for ChatGPT's responses to trigger finger questions ranged from 35 to 47, with an average of 42, indicating "moderate" quality. While 60% of the responses were satisfactory, 40% contained deficiencies. According to the JAMA Benchmark criteria, the absence of scientific references was a significant drawback. The average readability level corresponded to the university level, making the information difficult to understand for patients with low health literacy. Improvements are needed to enhance the accessibility and comprehensibility of the content for a broader patient population. To the best of our knowledge, this is the first study to investigate the use of ChatGPT in the context of trigger finger. While ChatGPT shows reasonable effectiveness in providing general information on trigger finger, expert oversight is necessary before it can be relied upon as a primary source for patient education.

People are increasingly using artificial intelligence (AI)-based chatbots to provide health-related information. However, concerns remain regarding the quality, accuracy, and readability of the information they produce. This study aimed to evaluate and compare the responses of five widely used AI chatbots to the most frequently searched keywords about apheresis. On May 1, 2025, the 25 most searched apheresis-related keywords were identified using Google Trends. Two keywords were excluded due to irrelevance. The remaining 23 queries were submitted to five chatbots: GPT-4o, Gemini 2.5, Grok 3, DeepSeek v3, and Copilot. Responses were assessed using the EQIP tool for content quality, the DISCERN questionnaire for information reliability, and the Flesch-Kincaid grade level (FKGL) and reading ease (FKRE) metrics for readability. Statistical analysis was performed using the Kruskal-Wallis test and Bonferroni correction. Significant differences were found among chatbots in EQIP, DISCERN, FKGL, and FKRE scores (p = 0.001). DeepSeek v3 demonstrated the highest quality and accuracy (EQIP: 95.7%, DISCERN: 71.8), while GPT-4o had the best readability (FKRE: 43.1, FKGL: 9.1). Copilot showed the poorest readability. Overall, chatbot responses were generally written at a college reading level. AI chatbots vary substantially in the quality and comprehensibility of their health information about apheresis. While newer models like DeepSeek offer improved informational accuracy, readability remains a concern across all platforms. Future chatbot development should prioritize plain-language communication to enhance accessibility and health literacy for diverse patient populations.

Artificial intelligence (AI) chatbots such as Chat Generative Pretrained Transformer-4 (ChatGPT-4) have made significant strides in generating human-like responses. Trained on an extensive corpus of medical literature, ChatGPT-4 has the potential to augment patient education materials. These chatbots may be beneficial to populations considering a diagnosis of colorectal cancer (CRC). However, the accuracy and quality of patient education materials are crucial for informed decision-making. Given workforce demands impacting holistic care, AI chatbots can bridge gaps in CRC information, reaching wider demographics and crossing language barriers. However, rigorous evaluation is essential to ensure accuracy, quality and readability. Therefore, this study aims to evaluate the efficacy, quality and readability of answers generated by ChatGPT-4 on CRC, utilizing patient-style question prompts. To evaluate ChatGPT-4, eight CRC-related questions were derived using peer-reviewed literature and Google Trends. Eight colorectal surgeons evaluated AI responses for accuracy, safety, appropriateness, actionability and effectiveness. Quality was assessed using validated tools: the Patient Education Materials Assessment Tool (PEMAT-AI), modified DISCERN (DISCERN-AI) and Global Quality Score (GQS). A number of readability assessments were measured including Flesch Reading Ease (FRE) and the Gunning Fog Index (GFI). The responses were generally accurate (median 4.00), safe (4.25), appropriate (4.00), actionable (4.00) and effective (4.00). Quality assessments rated PEMAT-AI as 'very good' (71.43), DISCERN-AI as 'fair' (12.00) and GQS as 'high' (4.00). Readability scores indicated difficulty (FRE 47.00, GFI 12.40), suggesting a higher educational level was required. This study concludes that ChatGPT-4 is capable of providing safe but nonspecific medical information, suggesting its potential as a patient education aid. However, enhancements in readability through contextual prompting and fine-tuning techniques are required before considering implementation into clinical practice.

ImportanceOnline patient education materials (PEMs) and large language model (LLM) outputs can provide critical health information for patients, yet their readability, quality, and reliability remain unclear for Meniere's disease.ObjectiveTo assess the readability, quality, and reliability of online PEMs and LLM-generated outputs on Meniere's disease.DesignCross-sectional study.SettingPEMs were identified from the first 40 Google Search results based on inclusion criteria. LLM outputs were extracted from unique interactions with ChatGPT and Google Gemini.ParticipantsThirty-one PEMs met inclusion criteria. LLM outputs were obtained from 3 unique interactions each with ChatGPT and Google Gemini.InterventionReadability was assessed using 5 validated formulas [Flesch Reading Ease (FRE), Flesch Kincaid Grade Level (FKGL), Gunning-Fog Index, Coleman-Liau Index, and Simple Measure of Gobbledygook Index]. Quality and reliability were assessed by 2 independent raters using the DISCERN tool.Main Outcome MeasuresReadability was assessed for adherence to the American Medical Association's (AMA) sixth-grade reading level guideline. Source reliability, as well as the completeness, accuracy, and clarity of treatment-related information, was evaluated using the DISCERN tool.ResultsThe most common PEM source type was academic institutions (32.2%), while the majority of PEMs (61.3%) originated from the United States. The mean FRE score for PEMs corresponded to a 10th- to 12th-grade reading level, whereas ChatGPT and Google Gemini outputs were classified at post-graduate and college reading levels, respectively. Only 16.1% of PEMs met the AMA's sixth-grade readability recommendation using the FKGL readability index, and no LLM outputs achieved this standard. Overall DISCERN scores categorized PEMs and ChatGPT outputs as "poor quality," while Google Gemini outputs were rated "fair quality." No significant differences were found in readability or DISCERN scores across PEM source types. Additionally, no significant correlation was identified between PEM readability, quality, and reliability scores.ConclusionsOnline PEMs and LLM-generated outputs on Meniere's disease do not meet AMA readability standards and are generally of poor quality and reliability.RelevanceFuture PEMs should prioritize improved readability while maintaining high-quality, reliable information to better support patient decision-making for patients with Meniere's disease.

Artificial intelligence (AI) models such as ChatGPT and DeepSeek have gained increasing attention for their potential to enhance patient education by delivering accessible and evidence-based health information. We designed the following study to evaluate the AI models-ChatGPT and DeepSeek-in generating patient education materials for bariatric surgery. Thirty commonly asked patient questions related to bariatric surgery were classified into four thematic domains: (1) surgical planning and technical considerations, (2) preoperative assessment and optimization, (3) postoperative care and complication management, and (4) long-term follow-up and disease management. Responses generated by ChatGPT and DeepSeek were evaluated using three key metrics: (1) response quality, assessed by the Global Quality Score, rated on a 5-point scale from 1 (poor) to 5 (excellent); (2) reliability, measured using modified DISCERN criteria, which assess adherence to clinical guidelines and evidence-based standards, with scores ranging from 5 (low) to 25 (high); and (3) readability, evaluated using two validated formulas: the Flesch-Kincaid Grade Level and the Flesch Reading Ease Score. ChatGPT significantly outperformed DeepSeek in response quality, with a median (IQR) Global Quality Score of 5.00 (4.00, 5.00) vs. 4.00 (4.00, 5.00) (P = 0.002). Higher reliability was also observed in ChatGPT, as reflected by mDISCERN scores across all four domains (median [IQR], 22.0 [21.0, 23.25] vs. 19.7 [19.0, 20.75]; P < 0.001). While no significant difference was found in the Flesch Reading Ease Score (mean [SD], 26.11 [12.84] vs. 20.87 [12.20]; P = 0.110), ChatGPT yielded significantly higher Flesch-Kincaid Grade Level Scores (meaning its text was more complex) (mean [SD], 16.40 [2.43] vs. 13.48 [2.35]; P < 0.001). Both models produced responses at a readability level corresponding to college education. ChatGPT provided higher-quality and more reliable responses, while DeepSeek's answers were slightly easier to read. However, both models' answers lacked attention to psychosocial and cultural aspects of patient care, highlighting the need for more empathetic, adaptive AI to support inclusive patient education.

The study assesses the quality, readability, reliability, and usefulness of exercise-related information generated by two large language models (LLMs), ChatGPT-4 and DeepSeek-V3, in response to frequently asked questions by patients with ankylosing spondylitis (AS). This cross-sectional comparative study developed a structured assessment framework using a set of exercise and rehabilitation-related questions, distributed across four key domains: exercise and physical activity (C1; 33 items), posture and mobility (C2; 6 items), breathing and pulmonary health (C3; 6 items), and general topics (C4; 5 items). Information quality was assessed using the modified DISCERN (mDISCERN) tool, while content reliability was evaluated with the Reliability Score and perceived usefulness was measured using the Usefulness Score. Readability was assessed using the Flesch Reading Ease (FRE) scale. Three independent physiotherapists with expertise in rheumatologic rehabilitation independently evaluated the responses. In total score comparisons, DeepSeek-V3 achieved significantly higher scores than ChatGPT-4 on the mDISCERN (4(3-4) vs. 3(3-3); p < 0.001), reliability (5(5-6) vs. 5(4-5); p < 0.001), and usefulness (6(5-6) vs. 5(5-6); p < 0.001). Domain-specific analysis showed higher usefulness scores for DeepSeek-V3 in C1 (p = 0.004), C2 (p = 0.019), and C4 (p = 0.005). Mean FRE scores were 30.4 ± 14.37 for ChatGPT-4 and 28.77 ± 17.77 for DeepSeek-V3, both classified as very difficult (p > 0.05). This study highlighted that responses generated by DeepSeek-V3 related to AS were generally more accurate and demonstrated greater reliability compared to those produced by ChatGPT-4. However, the complex language used by both LLMs may reduce accessibility for patients with limited health literacy. These limitations highlight the importance of healthcare professional oversight in exercise planning. Key Points • DeepSeek-V3 provided more accurate and reliable responses than ChatGPT-4 regarding exercise in AS. • Domain-specific analysis showed DeepSeek-V3 was particularly more useful in exercise, posture, and general topics. • Both LLMs generated content with very difficult readability, requiring college-level comprehension. • Healthcare professional supervision is essential when using LLMs in patient education.

Objective: This study aims to focus on the current application status of Artificial Intelligence Generated Content (AIGC) technology in the field of medical education, conduct an in-depth analysis of academic ethics issues among medical students in the context of the AI era, and propose strategies for academic norm construction, providing a reference basis for building and improving academic ethics construction among medical students. Methods: A cross-sectional survey was conducted among undergraduate medical students at a university in Zhejiang Province through questionnaires, collecting relevant data on medical students’ use of AIGC tools. SPSS 24.0 software was used for analysis. Results: Medical students have a strong opposition to the use of AIGC tools for fabricating and tampering with data and duplicating others’ academic achievements, but there are significant differences in attitudes, with some students exhibiting cognitive biases; they have a slightly higher tolerance for AIGC tools used to piece together papers and references. Medical students have a low understanding of professional content such as academic ethics management rules and academic misconduct investigation procedures; in the use of AIGC technology, 74% of medical students have AI applications on their devices, 80% have used AI tools to assist with academic tasks, and “occasionally use” accounts for 57%; 60% of students feel somewhat inconvenient but acceptable when unable to use AI, and feel less anxious. Conclusion: Medical students generally acknowledge that AIGC tools can enhance learning and research efficiency, yet there is a lack of effective recognition and proper guidance regarding their potential risks in practical applications. To address the severe challenges posed by AIGC technology to traditional higher education models, medical education needs to cultivate students’ correct academic ethics from the outset, regulate their research behavior, and make adjustments and reforms in the event of academic dishonesty. It is essential to nurture scientific research literacy that aligns with the standards of the academic community and maintains the healthy development of the medical research ecosystem.