Despite increasing calls for diversity across sectors (Teixeira et al., 2016), higher education curricula often lack systematic approaches to fostering diversity competence, although universities are both responsible for preparing students for heterogeneous work environments and are becoming more diverse themselves (Braun et al., 2020). There is a growing need for teaching formats that deepen students’ understanding of diversity while linking theoretical content to real world problems. Problem-based learning (PBL) has gained recognition for promoting critical thinking and professional skills through applied, student-centered learning (Brondani et al., 2024). This study examines how problem-based learning through hackathons contributes to the development of diversity competence in higher education. Thus, we ask the following research question: How does problem-based learning through hackathons contribute to the development of diversity competence in higher education? As part of the DiveHack project, a problem-based teaching model is conceptualized, implemented, and evaluated within a business and economics course at a German university. The study analyzes measurable changes in students’ perceived diversity competence and examines the consistency and development of learning outcomes across two consecutive summer semesters 2025 and 2026.

The theoretical framework draws on constructivist learning theory and the PBL learning cycle, which emphasizes student centered problem identification, hypothesis generation, and knowledge application (Hmelo-Silver, 2004). Hackathons represent a form of Short Term PBL that compresses intensive learning into 24 to 48 hours while maintaining core PBL principles of authentic, ill structured problems and collaborative learning (d’Escoffier & de Rezende, 2024). Diversity competence refers to the ability to work effectively across differences and is increasingly recognized as a key skill for navigating heterogeneous environments (Karagiannakis, 2024). By combining Short Term PBL with authentic diversity problems, the teaching model creates an applied learning environment in which students actively engage with complex organizational problems related to diversity management.

Methodologically, the study employs a sequential explanatory mixed methods design (Ivankova et al., 2006) implemented as a longitudinal trend study across two consecutive summer semesters. Longitudinal research enables the identification of changes, trends, or patterns within a population over time (Creswell, 2009). By repeating the same mixed methods sequence with independent cohorts, the study examines intervention consistency and replicability (Creswell & Plano Clark, 2018). The quantitative phase consists of pre and post surveys using Likert scale items adapted from established diversity competence frameworks (Braun et al., 2020). Statistical analyses include paired samples t tests within cohorts and independent samples t tests between cohorts. The qualitative phase involves focus group discussions and written reflections analyzed using qualitative content analysis (Mayring, 2012). The study follows established ethical standards with voluntary and anonymous participation (Cohen et al., 2017).

Initial findings from summer 2025 demonstrate measurable increases in perceived diversity competence. Pre survey data revealed that 73 percent of students reported low prior knowledge of diversity management before the intervention. Post survey results showed that 100 percent of participants found the hackathon helpful for understanding diversity, with 81.8 percent rating it as very helpful. All participants expressed satisfaction with the hackathon format. Qualitative findings indicate that students particularly valued the practical application of theoretical knowledge, interdisciplinary collaboration, and engagement with authentic organizational challenges.

The study is subject to several limitations, including reliance on self-assessment measures, relatively small sample sizes at a single institution, and the focus on immediate post intervention effects rather than long term retention. Longitudinal mixed methods research is motivated by the analytical interest in documenting how attitudes or experiences change over time and isolating key variables across different data sweeps (Fielding, 2018). In addition, the trend study design with independent cohorts does not allow conclusions about individual level development over time (Creswell, 2009). Future research should incorporate objective measures of diversity competence and extended follow up assessments.

Overall, hackathon based PBL formats provide a structured opportunity to bridge the theory practice gap through authentic diversity challenges. This approach supports the development of diversity competence and applied problem-solving skills (Brondani et al., 2024). The findings indicate that such formats represent a viable model for integrating diversity education into higher education curricula (Braun et al., 2020) and provide empirical support for Short Term PBL in fostering measurable diversity competence development (d’Escoffier & de Rezende, 2024).

Teaching chemistry and other natural sciences at the university level is more resource‑intensive than teaching social sciences or humanities, as it requires specialised infrastructure such as laboratories, instrumentation, and safety systems. While laboratory work may appear to simplify instruction, it also demands that students develop specific practical skills.

Despite these differences, chemistry curricula are broadly comparable to those of other disciplines in terms of teaching methods and assessment. Although the content and learning objectives are discipline‑specific, the pedagogical framework is similar, and—as in all fields—the quality of instructor–student interaction remains crucial for student motivation and effective learning [1].

According to the predictive models proposed by Held and Mori [1], teachers’ support for student autonomy, learning processes, and instructional design plays a crucial role in fostering student motivation. Among these factors, support for learning exhibits the highest predictive value, with an expectancy of 0.91. The remaining forms of support show similarly high expectancies, at approximately 0.85. Another important determinant is teachers’ error‑management behavior, which demonstrates a comparable expectancy value of 0.84.

Science and education policy in chemistry is defined through official national regulatory documents as well as through guidelines issued by professional chemical societies, such as the American Chemical Society [2] and the Royal Society of Chemistry [3]. In Lithuania, chemistry education is similarly governed by a nationally approved curriculum issued by the Ministry of Education, Science and Sport [4].

One of the most fundamental principles emphasized across all chemistry education policies is laboratory safety [5]. Students are not permitted to undertake practical laboratory work without prior approval and supervision by an instructor. In addition to safety, chemistry curricula universally emphasize a balanced integration of theoretical knowledge and practical competence [3–5]. The theoretical curriculum typically covers atomic theory, chemical bonding, molecular structure, thermodynamics, kinetics, chemical equilibrium, and quantitative reasoning. The practical component focuses on experimental design, data interpretation, and ethical scientific practice. Furthermore, communicating chemical knowledge in both written and oral forms is recognized as a critical learning outcome across chemistry education policies [3–5].

In this context, chemistry education promotes analytical thinking by engaging students in data interpretation, experimental design, and evidence‑based reasoning. For example, Doughan and Shahmuradyan report that inquiry‑driven laboratory courses, in which students design and refine their own experiments, enhance critical thinking, creativity, and scientific communication skills [6]. Creativity is also central to chemistry learning, as students must visualize molecular processes, explore alternative solution pathways, and troubleshoot unexpected experimental results [7].

Students enter university with prior knowledge acquired during school education, which must be critically examined, expanded, and refined in higher education. As emphasized by Taber [8], “university teachers with some understanding of how this knowledge base has developed are best able to help students to build on it effectively.” Consequently, teaching strategies should explicitly acknowledge and build upon students’ pre‑existing conceptions. Extensive research demonstrates that active‑learning approaches significantly enhance student achievement in undergraduate STEM courses when compared with traditional lecture‑based instruction, including in large‑enrollment chemistry classes. These effects are particularly pronounced in traditionally challenging subjects, such as organic chemistry, where active‑learning strategies have led to measurable improvements in student success and retention [9,10].

In the era of artificial intelligence (AI), an ongoing discussion concerns the legality, appropriateness, and extent of AI use in higher education in chemistry. In particular, questions have arisen regarding how AI tools may be integrated into learning and assessment without compromising academic integrity. Erümit and Sarıalioğlu [11] address these issues by examining ethical considerations related to the use of AI in science and chemistry education. They emphasize the importance of increasing both student and instructor awareness of responsible and informed AI use to foster ethical decision‑making and promote the conscious, transparent application of AI tools in educational contexts.

Although the curriculum for chemistry education is well established, ongoing changes in the global environment, rapid scientific innovation, and technological progress necessitate the continuous development of new pedagogical approaches and the incorporation of emerging perspectives. These evolving challenges will be presented as a review and discussed further in the EAIR forum.

This study investigates how AI-generated, thinking-oriented feedback supports far transfer through asynchronous group learning in university education. It addresses three research questions:
(1) How are disciplinary concepts reinterpreted when applied to qualitatively different domains?
(2) What role does AI-mediated feedback play in this reinterpretation process?
(3) How does asynchronous group interaction influence the social construction of transfer?
　Far transfer is widely recognized as a central objective of university education, particularly in contexts where higher education is expected to produce socially relevant and publicly meaningful knowledge. Yet empirical research consistently shows that disciplinary knowledge often remains confined to its original epistemic frame and is seldom mobilized across domains. This gap raises concerns about how universities can demonstrate the broader value of academic learning and sustain trust with society.
　To address this issue, the present study reconceptualizes far transfer as a socially mediated, dialogic process rather than an individual cognitive outcome. The research examines an undergraduate course in agricultural economics in which the interpretation of artworks was explicitly designed as a transfer task. Students were asked to apply core disciplinary concepts—such as efficiency, externalities, sustainability, and value evaluation—to analyze artworks and artistic expression.
Generative AI was embedded in the course as a feedback mediator. Rather than providing evaluative judgments or model answers, the system generated interpretive comments and open-ended questions based on students’ written reports and discussion posts. The feedback aimed to externalize conceptual assumptions and prompt reconsideration from alternative disciplinary or societal perspectives. In this design, AI functioned not as an authority but as a catalyst for reflective inquiry.
　Group learning was conducted through asynchronous discussions on Blackboard Learn. Each group engaged in iterative written exchanges, using AI-generated feedback as a common reference point. The asynchronous format allowed students to revisit peers’ interpretations, critically assess conceptual applications, and refine their reasoning over time. Written interaction logs created a transparent epistemic space in which disciplinary assumptions could be articulated, scrutinized, and collectively evaluated.
　The data corpus included students’ individual reports, full transcripts of asynchronous group discussions, and AI-generated feedback logs. A qualitative, process-oriented analysis was undertaken. Posts were segmented into meaning units, and conceptual shifts were traced across interactional sequences. Particular attention was given to moments where concepts were reframed, contested, or abandoned. This analytical framework draws on research in collaborative learning and socially situated knowledge construction.
　Findings indicate that AI-generated feedback did not steer students toward predetermined interpretations. Instead, it functioned as a generative starting point that facilitated the experimental reuse of disciplinary concepts. The combination of AI-mediated prompts and asynchronous peer dialogue enabled students to detach concepts from their original context and evaluate their relevance to non-economic phenomena. Transfer emerged gradually through negotiation, revision, and collective sense-making.
　This study contributes to higher education research in three respects. First, it provides empirical evidence that far transfer can be observed and analyzed as an interactional process rather than inferred solely from outcome measures. Second, it clarifies how generative AI can operate as a pedagogical infrastructure that enhances epistemic transparency and dialogic engagement. Third, it offers design principles for courses seeking to strengthen the societal relevance of disciplinary learning and reinforce public trust in higher education’s capacity to cultivate transferable understanding.
　By reframing AI-supported transfer as a collaborative evaluative process, the study contributes to ongoing EAIR discussions on teaching, learning, and evaluation practices that foreground the public value of university education.