Understanding Cognitive Debt Accumulation When Utilizing AI Assistants for Essay Writing

This study examines the cognitive costs of using Large Language Models (LLMs), such as ChatGPT, in an educational context, specifically for essay writing. Participants were divided into three groups—LLM, Search Engine, and Brain-only—across four writing sessions. EEG recordings tracked their brain activity to evaluate cognitive engagement. Results indicated varied neural connectivity patterns among groups: the Brain-only group displayed the strongest networks, while the LLM group showed the weakest engagement. In the fourth session, where groups switched tools, participants who previously used the LLM struggled with recalling details from their essays. Overall, the LLM group’s essays showed lower ownership compared to the Search Engine and Brain-only groups, with diminishing learning skills observed over time. The findings highlight the significant impact of LLMs on cognitive functioning and learning, suggesting a need for caution in their integration into educational settings.

Source link