This report provides a comprehensive analysis of the current trends, policies, and practices related to the use of Large Language Models (LLMs) and Generative AI in higher education across the partner institutions involved in the ADMIT project. The findings are based on qualitative and quantitative data collected from key stakeholders, including students, teachers, IT/support staff, and administrators. The study reveals significant variability in awareness, application, and institutional strategies for leveraging AI in education. At the individual level, awareness of AI technologies varies widely among students, teachers, and IT/support staff. Teachers and students primarily rely on self-directed learning, engaging with online resources and experimenting with tools like ChatGPT. IT/support staff, often responsible for piloting technical projects, demonstrate slightly higher familiarity. However, the absence of structured training or institutional support leaves knowledge inconsistent across individuals. AI is predominantly used experimentally by individuals. Students employ it for brainstorming, refining academic writing, and generating ideas. Teachers use it for research, creating instructional content, quizzes, and assessments. IT/support staff explore its use for troubleshooting and piloting small-scale educational tools. These applications indicate that AI adoption remains in an exploratory phase at the individual level. AI offers opportunities to enhance productivity, reduce repetitive tasks, and provide personalized learning experiences. However, individuals express concerns about ethical dilemmas, overreliance, and the potential erosion of critical thinking skills. For IT/support staff, data privacy and security are prominent challenges that complicate adoption and scalability. Awareness at the institutional level is fragmented, often driven by specific departments or proactive faculty members. Students are less familiar with institutional policies and practices compared to teachers, IT/support staff, and administrators. While some universities host informal workshops and discussions, most lack cohesive policies to promote awareness and encourage widespread adoption of AI. This inconsistency highlights the need for structured institutional strategies. AI applications at the institutional level are primarily limited to pilot projects or niche use cases, such as AI-powered chatbots for student support, Moodle plugins for teaching, and automated grading systems. While these efforts demonstrate the potential of AI, financial, technical, and organizational barriers hinder their scalability across institutions. Institutions see the potential for AI to transform teaching, learning, and administrative functions. However, challenges include staff resistance to adopting new technologies, ethical concerns around AI usage, and limited resources. Data privacy and equitable access to AI tools are additional hurdles that institutions must address. National policies often reflect a general awareness of AI’s transformative potential but lack specificity in addressing educational needs. Institutions frequently cite the absence of actionable, education-specific guidelines as a significant barrier to aligning with national strategies. National-level support for AI remains underdeveloped, focusing on ethical considerations and data security rather than practical implementation. While some pilot projects receive backing, institutions are often left to navigate AI integration independently. National policies could foster equitable access, standardize best practices, and create collaborative opportunities among institutions. However, insufficient funding, inconsistent guidelines, and inadequate infrastructure remain critical challenges at the national level.

ei tietoa saavutettavuudesta

unknown accessibility

nonPeerReviewed