Students’ conceptions of artificial intelligence and machine learning in K–12 education: a systematic review

Background and context Knowing students’ conceptions of computer science phenomena is essential for the design of teaching–learning settings and instruments. They are particularly interesting in teaching artificial intelligence (AI) and machine learning (ML) concepts. Therefore, a comprehensive synthesis focused on K–12 is essential to provide educators with actionable conceptions for curriculum design and assessment.Objective We investigate common conceptions of AI/ML in K–12 and to which curricular concepts these can be assigned. We also examine how conceptions vary across age/grade groups and the data collection/evaluation methods employed. This study uniquely synthesizes fragmented research on students’ conceptions of AI/ML into a unified snapshot, offering a systematic mapping that informs curriculum recommendations.Method A systematic literature review according to PRISMA is used to identify the relevant studies. We analyzed the selected literature, extracted the student conceptions, and classified them according to the curricular concepts.Findings We identified 27 relevant studies (from 2007 to 2024) including the conceptions of 1,077 K–12 students from 30 different countries. In addition, our data includes further information on the type of school, age/grade levels, genders and nationalities, the availability of the collected data, and the survey and evaluation methods used. We took a closer look at 29 of the 409 resulting, highly prevalent conceptions, which stand out due to their frequent occurrence.Implications Our findings may be useful for determining in which grade levels interventions targeting specific concepts would be most effective. This could also be used to study the development of these conceptions or developing assessments.