Cecilia Ka Yuk Chan

Research funding bodies have significant influence on what research is prioritised, how it is conducted, and what standards are enforced. While a number of global frameworks provide normative guidance on AI ethics, there is a clear gap in understanding how these principles are implemented at the funding-body level. This study aims to examine the current AI policies and guidelines of research funding bodies across different geographical regions. Inductive thematic analysis was adopted to analyse 56 AI-related policies from 46 funding agencies across the US, the UK, Europe, Australia, Singapore, Mainland China, and Hong Kong. The analysis revealed five common themes: human-centredness, safeguarding, embeddedness, governance, and AI readiness. It was also found that current funding-specific policies lack explicit guidelines on how broad principles and macro-level regulations can be interpreted and applied in AI-enabled research funding processes. The findings underscore the important role of funding agencies as key actors in defining responsible AI use in research processes. Finally, the practical implications of this study for regulators, funding bodies, and researchers are discussed, highlighting the value of policy document analysis in shaping ethical, transparent, and responsible AI frameworks.

Background: The integration of Generative AI (GenAI) into collaborative learning in secondary education has created new possibilities for this pedagogical approach. It has also raised concerns about its impact on team communication, particularly, for adolescent learners whose collaborative competencies are still developing. Current research offers limited understanding of how GenAI shapes team communication in educational settings. Objectives: This study explores how GenAI influences team communication during group tasks and identifies key factors shaping these dynamics in secondary education. Methods: A mixed‐methods design was employed with 24 secondary students (Grades 9–12) from two Hong Kong secondary schools. Data collection included written reflections from all participants and semistructured interviews with 10 students. Thematic analysis addressed GenAI's impacts on communication frequency, timeliness and quality (RQ1), while activity theory guided analysis of influencing factors (RQ2). Results and Conclusions: Findings underscore GenAI's dual role as both a facilitator and a potential disruptor of team interaction in secondary education. First, the introduction of GenAI tools improves communication quality by fostering deeper understanding, generating new ideas and resolving disagreements. However, generative AI tools also significantly reduce communication frequency among team members and may introduce delays in team exchanges. Moreover, this study identifies six factors influencing the use of GenAI in team communication as perceived by students, including the perceived value of discussions, task interdependence, team familiarity, attitudes towards AI, guidelines for AI use and role clarity.

The rapid integration of large language models (LLMs) into education has shifted research focus from questions of capability, such as what LLMs can do and how accurately—to questions of deployability, including how they can be operated effectively for many learners at once. In school environments, system reliability, scalability, and real-time responsiveness are critical, as delays or interruptions can directly reduce learner engagement, particularly during synchronous activities. This study evaluates the performance of open-source LLaMA models ranging from 1 billion to 70 billion parameters across one-, dual-, triple-, and quad-GPU configurations suitable for educational settings. Performance is assessed using four key indicators: success rate (percentage of completed requests), generation speed (tokens per second), throughput (completed responses per second), and latency (time until full response generation). These metrics were measured under progressively increasing numbers of simultaneous users to identify system capacity limits and trade-offs between model size, responsiveness, and scalability. The results indicate that smaller models (1B–3B) deliver faster, more stable performance under concurrent use, while larger models (8B–70B) experience substantial slowdowns and reduced reliability, even on high-end GPU systems. These findings suggest that effective educational deployment should prioritize empirical performance and infrastructure compatibility over model size alone. The paper concludes by proposing a practical framework to guide educators, administrators, and developers in selecting and configuring locally hosted GPU systems that balance model capability, response speed, and resource efficiency for real-time applications such as AI tutors, classroom chatbots, and automated feedback tools.

With the emergence of entrepreneurship education, its integration within the arts discipline remains limited. Existing literature on entrepreneurship education in higher education has primarily focused on outcomes, with assessments relatively overlooked. This study investigated entrepreneurship education assessment through the expectations and perceptions of multiple stakeholders, including students, teachers and professionals from the field of arts. An open‐ended survey was administered to first and final year students at a university in Hong Kong, followed by individual semi‐structured interviews with higher education students and educators, as well as professionals. Findings showed that entrepreneurship education was valued in the arts discipline, and results recommended focusing on creative processes and fostering entrepreneurial competencies to enhance employability and career management. Non‐traditional but authentic and experiential approaches in evaluation methods were highlighted. In light of the findings, this study offers valuable insights to curriculum design, as well as teaching and learning in the arts discipline within higher education.

This study investigates secondary school students' perceptions of generative AI (GenAI) as virtual mental health therapists within educational settings. Leveraging data from the written reflections of 69 Hong Kong secondary school students engaged in an AI literacy programme, the study applied inductive thematic analysis to explore perceived potentials and limitations of GenAI therapy chatbots. Students valued the accessibility, anonymity and cost‐effectiveness of AI‐powered therapy. However, significant concerns arose regarding the technology's lack of genuine empathy, limited understanding of non‐verbal cues and potential for inaccurate responses. Ethical issues, such as data privacy and trust, were highlighted, with students expressing warranted distrust regarding personal data handling. The findings emphasised the risk of over‐reliance on AI and the therapeutic misconception that AI generates more suitable advice. While students acknowledged GenAI as a supplementary tool for initial support, they underscored the irreplaceable role of human therapists for deep, personalised care. This research calls for comprehensive AI literacy education to help students effectively navigate the strengths and boundaries of AI in mental health support.

This study develops a human-centered design (HCD) approach to create a GenAI trainer that addresses critical gaps in virtual teamwork training for engineering students. While virtual teamwork competency is increasingly essential, current programs often prioritize task completion over competency development. Leveraging generative AI's capabilities for personalized interaction, scenario simulation, and tailored feedback, we employ a three-phase HCD approach: (1) identifying unmet user needs through stakeholder interviews, revealing key challenges in instructional support, training formats, feedback mechanisms, and teamwork awareness; (2) co-designing solutions with instructors and students to create an AI trainer that combines Socratic questioning and scenario-based learning; and (3) testing the system and obtaining feedback from engineering students. Results demonstrate significant improvements across multiple dimensions: transforming passive learning into active experiences, delivering real-time actionable feedback, enhancing conceptual understanding and awareness of virtual teamwork, and developing practical virtual teamwork skills through authentic scenarios. Participant feedback also identified future improvements for enhanced personalization and immersion. This study contributes both theoretically and practically by illustrating how HCD can effectively integrate AI capabilities with pedagogical needs, while providing a replicable model for developing competency-based training tools that balance technological innovation with educational effectiveness.