Can we and should we use artificial intelligence for formative assessment in science?

In this commentary, we respond to the recent article Applying machine learning to automatically assess scientific models by Zhai et al. (2022).The authors present automated assessment as a solution to the problem of limited time for assessment in middle school science classrooms.Drawing from our collective expertise in science assessment, machine learning (ML), artificial intelligence (AI), and culturally relevant and linguistically responsive pedagogy, we argue that there are significant limitations to the current applications of AI for formative assessment practices.Although we believe that these limitations extend to all students, we are particularly concerned about the implications for students from nondominant cultural and linguistic backgrounds.We first share our understanding of AI's role in formative assessment, with reference to the paper by Zhai and colleagues.Next, we ask whether AI can effectively assess students' emergent sensemaking and then consider whether we should use AI for purposes of formative assessment.Finally, we discuss how we can better use AI for formative assessment. | ARTIFICIAL INTELLIGENCE AND ASSESSMENT PRACTICESML is a form of AI that uses datasets to train algorithms to identify patterns and relationships, which can be used to simulate human decision-making.Zhai et al. (2021) argue that ML can All authors have made equal and significant contributions to this paper.