Study finds LLMs provide high levels of assistance in AI tutoring

A new study published on arXiv analyzes the pedagogical behavior of Large Language Models (LLMs) when used as AI tutors. Researchers developed a five-level scale to measure the degree of direct assistance provided by LLMs, finding that over 95% of responses in a university AI course were highly direct, categorized as 'Explaining' or 'Solving'. While this level of assistance correlated with subsequent student conversational behavior, it offered little additional predictive power for exam performance beyond prior achievement and dialogue patterns. The findings establish an empirical baseline for LLM tutoring and offer a framework for evaluating future tutoring designs. AI

IMPACT Establishes a framework for evaluating LLM tutoring effectiveness and provides insights into their current assistance levels in educational settings.

RANK_REASON The cluster contains a research paper published on arXiv detailing findings about LLM behavior in an educational context. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• alphaXiv
• arXiv
• CatalyzeX
• Connected Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• Litmaps
• ScienceCast
• scite Smart Citations

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →