A reconceptualisation of the instrumental genesis process in the context of human–LLM interactions

In this chapter, we discuss the role that Large Language Models could play as tools for student self-assessment, reflecting on the complexity that the use of these tools brings to the analysis of the interaction between humans and digital resources. In particular, based on the results of a study focusing on students’ perspectives on the role that GPT-4o could play in supporting self-assessment in mathematics, we propose a reconceptualisation of the instrumental genesis process developed to take into account this complexity, highlighting the main new elements that the human-LLM interaction brings to the process itself. We complement our reconceptualisation by introducing a metaphor that represents instrumental genesis as noise reduction.