ABSTRACT. This study examines exploratory learning activities in the context of programming a neural network for classification. Building on a concise starter project, it identi-fies activities aligned with Kolb’s model of experiential learning. These activities can be initiated through targeted instructional prompts and carried out inde-pendently by students, supported by generative AI tools.

ABSTRACT. As Artificial Intelligence (AI) continues to evolve and influence education, global discussions increasingly call for curriculum reforms that integrate AI literacy and AI competencies. To better understand how educational jurisdic-tions are responding to this change, a literature search was conducted to ex-plore educational policy guidelines and the extent to which national curricu-la focus on Artificial Intelligence (AI) and related competencies. To guide this review, five AI in education (AIED) frameworks were selected for com-parative analysis: the AI4K12 framework (USA), the AI Literacy Framework (EU), the Chinese National Guidelines for AI Education (China), the Aus-tralian Generative AI in Schools (Australia) and the AI Competency Frame-work for Students (UNESCO). This paper presents a comparison of these global five frameworks, through document and thematic analysis, examining their underlying goals, objectives, focus areas, pedagogical perspectives and concepts covered. The analysis highlights the similarities and differences centred around the themes that were derived from AI literature. The findings revealed significant alignment across the five frameworks on key themes such as the ethical use of AI, AI literacy, and skill development in designing AI solutions. However, notable differences were observed in their specific educational objectives, thematic focus, and the range of AI topics addressed. These findings can provide valuable insight for AI curriculum development, guiding educators and policymakers in creating balanced and comprehensive AI education programs.

ABSTRACT. This study explores German 5th-grade students’ conceptions of artificial intelligence (AI) and the factors influencing them. Using a modified questionnaire and qualitative content analysis, responses from 48 students were examined. Results reveal that nearly half of the participants either lacked a concept of AI or misunderstood it, often associating AI with creativity due to linguistic confusion between “künstlich” (artificial) and “künstlerisch” (artistic). Although 75% were familiar with voice assistants, very few identified these as AI systems. Media -especially television, films, and social media - was the most frequently cited influence, often fostering anthropomorphic or robotic conceptions of AI. Additional influences included family, teachers, and everyday experiences. The findings emphasize the need for targeted AI education that addresses prevalent misconceptions and leverages students’ lived experiences to build accurate mental models.

ABSTRACT. Agile methods from software development have increasingly been adopted into K-12 computer science education. In this position paper, we explore how lessons learned from industry about the distinction between simply applying agile practices (Doing Agile) and genuinely embedding agile values (Being Agile) can enrich K-12 computer science education. Our literature review shows generally positive impacts of various agile-based educational approaches on student engagement and project outcomes. However, explicit teaching and sustained integration of agile values have so far received little to no attention in research.We argue that explicitly fostering an agile mindset with focus on adaptability, collaboration, and continuous reflection significantly enhances computer science education, equipping students to navigate a complex and uncertain future. Finally, we outline essential conditions for successfully embedding agile values and highlight critical areas requiring further empirical research.

ABSTRACT. Agentic AI creates profound contradictions in higher education, challenging traditional pedagogy. This symposium explores ways to resolve these contradictions and expand learning environments. Uniting perspectives on ethics, authentic assessment, and human-AI agency, it fosters dialogue on redesigning pedagogy for authentic learning, centred on human dignity.

Presenters:

• Yuko Murakami – Humanities: What Should Be Nurtured in the Age of GAI

• Said A.S. Yunus & Eliana El-Khoury – Deepening Formative and Alternative Assessment of Human Skills in AI-Augmented Learning Environments

• Mary Webb – AI-Enhanced Formative Assessment in Higher Education: Opportunities and Challenges

ABSTRACT. Over the past 25 years, Design-Based Research (DBR) has progressively established itself as a promising methodological paradigm for educational technology. Emerging in the early 2000s and grounded in pragmatic constructivism, DBR combines the design of educational artefacts with the production of scientific knowledge through iterative, collaborative, and context-sensitive processes. This paper offers a critical overview of DBR’s status in the field of educational research, with a particular focus on technology-enhanced learning. It contributes to the field by (1) clarifying DBR’s epistemological foundations and methodological distinctiveness, (2) analyzing the types of knowledge it produces, including practical, design-oriented, and theoretical knowledge, and (3) illustrating these dimensions through the PLAY project, a multi-year DBR initiative centered on game-based learning in museum settings. The paper also identifies key challenges facing DBR today, particularly in balancing scientific rigor with practical relevance, and offers a set of recommendations aimed at consolidating DBR as a scientifically grounded mode of educational inquiry, capable of generating meaningful contributions to the field of research on digital technology in education and training.

ABSTRACT. After a brief history of more than 70 years of research in the Artificial Intel-ligence field, this paper sheds light on the different categories of AI tech-niques and compare them with Deep Learning across explainability, energy consumption and reliability. While scientific literature already established the dramatic carbon footprint of GAI tools, this research explores HE teacher’s values and position concerning this ecological impact by the mean of semi-directed interviews. After a categorization phase, the data are analyzed ac-cording a quantitative method to represent the part of ecology compared with the other values they are concerned with. A qualitative analysis brings the contributions of the participants on this ecological aspect and reveals four typical profiles. This study shows that most of the participants tend to ignore this ecological problem when teaching GAI use. The main message of this contribution is to warn the community on the Educators’ responsibility in GAI tools proliferation.

ABSTRACT. This paper examines the prominence and influence of breakthrough technologies (BTs) in current educational research, focusing on publications from leading journals in the Education & Educational Research category and the official journal of IFIP's Technical Committee 3 (TC3). The study analyses the extent to which BTs are represented in top-tier journals, highlighting the significant role of AI-related technologies and their transformative potential in education. The paper also discusses the alignment of IFIP TC3's activities and its journal with international research trends, emphasizing the committee's commitment to addressing challenges posed by BTs in education. Findings suggest that BTs are not a passing fad but a solid trend shaping the future educational landscape. However, the paper calls for deeper theoretical grounding and rigorous research to fully harness these technologies' potential.

ABSTRACT. The consideration of using generative AI is being advanced across various domains. In programming education, however, there are concerns regarding its implementation, including the possibility that presenting program code may prevent students from thinking and the difficulty of describing accurate assignment specifications into prompts. To address these issues, this study designed and developed "ChotGPT", which provides functions to "avoid presenting direct solutions" and "assist generating prompt for code review using assignment specifications", with the aim of supporting learning in introductory programming courses. An empirical study has been conducted with approximately 280 first-year students in an introductory programming course in an interdisciplinary department during the 2024 academic year. The average usage rate per assignment reached approximately 30%. Through analysis of usage logs and questionnaires, we checked that the effectiveness of the two distinctive functions were generally supported by students throughout the exercises over ten weeks.