Secondary school students' conceptions of learning AI range from basic awareness and knowledge acquisition to sophisticated views involving ethical considerations and transformative intellectual development, with most students demonstrating positive attitudes but fewer reaching advanced competency levels.
Objective: The main goal of this phenomenographic study was to investigate secondary school students' conceptions of learning artificial intelligence (AI) in order to understand how students perceive and engage with AI education. The research aimed to identify different categories of students' conceptions and explore the hierarchical relationships between these conceptions to inform curriculum development and teaching practices in K-12 AI education.
Methods: This qualitative phenomenographic study involved 88 secondary school students (56.8% male) from Hong Kong who participated in focus group interviews after completing an AI curriculum. The study was conducted in two batches, with students from grades 7-11 across 16 publicly funded secondary schools representing diverse socio-economic backgrounds. Data was collected through semi-structured interviews using six key questions exploring students' motivations, experiences, learning processes, and understanding of AI. The AI curriculum covered three main aspects: knowledge of AI, processes in AI, and impacts of AI, with five modules per chapter (Awareness, Knowledge, Ethics, Interaction, and Empowerment). Data analysis followed phenomenographic methodology using iterative coding processes, with interrater reliability calculated using Cohen's kappa (0.87). Categories of description and hierarchical outcome space were developed to represent the range of students' conceptions.
Key Findings: Six hierarchical categories of students' conceptions of learning AI were identified: (1) Gaining Awareness - recognizing AI technologies in daily life (70.4% of students); (2) Acquiring Knowledge - understanding fundamental AI concepts and mechanisms (79.5%); (3) Satisfying Interest - developing curiosity and enthusiasm for AI through hands-on activities (75%); (4) Considering Social Impacts - emphasizing ethical implications and societal responsibilities (44.3% achieved this level); (5) Improving Self-competency - focusing on practical applications and problem-solving capabilities (18.2%); and (6) Seeing in a New Way - viewing AI as transforming general cognitive abilities and critical thinking skills (10.2%). The outcome space revealed transitions across four dimensions: conception domain (knowledge→attitude→competency), conception schema (reproductive→constructivist), learning strategies (surface→deep), and motivation orientation (extrinsic→intrinsic). Students generally held positive attitudes toward AI learning, with most focusing on daily applications and basic knowledge acquisition, while fewer reached advanced competency levels involving creative application and ethical reasoning.
Implications: The findings contribute significantly to understanding how students conceptualize AI learning and provide crucial insights for curriculum development in K-12 AI education. The study reveals that while most students achieve basic awareness and knowledge levels, fewer progress to advanced competency stages, suggesting the need for more carefully designed curricula that scaffold students toward higher-order thinking. The research highlights the importance of integrating AI ethics education, as students demonstrated strong awareness of social impacts. The hierarchical nature of conceptions suggests that early-stage learners benefit from experiential activities fostering awareness and interest, while advanced learners require opportunities for meaningful application and ethical reflection. The study supports the necessity of interdisciplinary AI education that connects to students' daily lives and demonstrates practical relevance. These insights inform educators about aligning teaching practices with students' evolving conceptions and developing age-appropriate, ethically informed AI education programs.
Limitations: The study has several methodological and contextual limitations. From a methodological standpoint, the reliance on interview data may be subject to social desirability bias, where students provided responses they believed were expected rather than their genuine beliefs. Future research should triangulate interview data with more objective sources such as classroom observations or student artifacts. The study was conducted exclusively with Hong Kong secondary school students, which may limit transferability to other cultural contexts or educational systems. The sample, while representing diverse socio-economic backgrounds within Hong Kong, may not accurately represent other learner populations. Additionally, the qualitative nature of phenomenographic research means findings focus on understanding rather than statistical generalizability.
Future Directions: The study recommends several areas for future research to expand understanding of AI learning conceptions. Replication studies across broader cultural contexts and different school settings (including private schools) would enhance generalizability of findings. Longitudinal research tracking students' conception development over time could illuminate how AI learning experiences evolve and identify critical transition points. Comparative studies examining conceptions across different AI curriculum frameworks would help identify most effective pedagogical approaches. Integration of multiple research methods, combining phenomenographic interviews with drawing-based methods, classroom observations, and learning analytics, could provide more comprehensive understanding of students' AI learning experiences. Research investigating the relationship between students' conceptions and actual learning outcomes, as well as studies exploring teacher conceptions and their impact on student learning, would further advance the field.
Title and Authors: "A phenomenographic approach to students' conceptions of learning artificial intelligence (AI) in secondary schools" by Tianle Dong, King Woon Yau, Ching Sing Chai, Thomas K.F. Chiu, Helen Meng, Irwin King, Savio W.H. Wong, and Yeung Yam.
Published On: September 8, 2025 (online publication)
Published By: Education and Information Technologies (Springer)