Cognitive Debt Accumulation in Essay Writing with AI Assistants

This study investigates the effects of LLM-assisted essay writing on neural and behavioral aspects. Participants were split into three groups: LLM, Search Engine, and Brain-only (no tools), completing three sessions each, followed by a fourth session with group reassignment. A total of 54 participants were involved, with 18 completing all sessions. Electroencephalography (EEG) measured cognitive load, while essays were analyzed using Natural Language Processing (NLP) and scored by human teachers and an AI judge. Results indicated significant differences in brain connectivity, with Brain-only users showing the strongest networks, while LLM users exhibited the weakest. In the final session, LLM-to-Brain participants showed decreased neural engagement, whereas Brain-to-LLM users demonstrated improved memory recall. Ownership of essays was lowest among LLM users, who also struggled with quoting their work. Findings suggest that despite the convenience of LLMs, they may incur cognitive costs, raising concerns about their long-term educational impact and the role of AI in learning.

Source link