Is ChatGPT Massively Used by Students Nowadays? A Survey on the Use of Large Language Models such as ChatGPT in Educational Settings

Generative AI Tools (LLMs, AI Assistants)

Context: Educational settings across secondary school, high school, and university levels in France and Italy, also applicable to students with learning disabilities, and general curriculum adaptation. Encompasses a wide range of subjects and educational levels.

Implementation: Use of generative AI tools (e.g., ChatGPT, customized AI assistants) for various educational purposes. This includes student use of LLMs for content generation, and curriculum adaptation to incorporate AI tools. The implementation involves surveys to gauge usage, and hypothetical and practical integrations of AI tools into curricula and homework design. It also includes suggesting changes in homework design to account for tools like ChatGPT as a potential support.

Outcomes: Widespread use across age groups and disciplines. Enhanced learning through customized programs and AI assistants, particularly for students with learning disabilities. Emphasizing subject appropriation, fundamental understanding, critical thinking, and memorization.

Challenges: Potential risks to cognitive skills development (critical thinking, foundational knowledge). Unsupervised and extensive use may lead to unintended consequences, such as the decline of essential cognitive skills or a weaker foundation in basic knowledge. Requires careful integration and management to avoid over-reliance and negative impacts. Need to teach students to critically evaluate and revise outputs from these tools. Potential for unintended consequences and the need for reevaluation of pedagogical strategies.

Ethical/Educational

Over-reliance on AI tools, potentially eroding critical thinking and foundational cognitive skills. Potential for erosion of critical thinking and foundational cognitive skills due to over-reliance on AI for tasks like summarization and basic logic.

Proposed Solutions: Teaching students to critically evaluate and revise outputs from AI tools, integrating these systems into teaching methods, focusing on ethics, critical thinking, and limitations of AI. Revising teaching and evaluation methods to account for LLMs, emphasizing subject appropriation, fundamental understanding, critical thinking, and memorization rather than raw answers. Designing homework to include unassisted restitution parts for students to appropriate the answers produced by these tools and in some case mobilize memorization skills.

Gender Disparity

Emerging AI gap between male and female students, with males exhibiting higher usage rates and a greater tendency to systematically revise generated responses, suggesting a higher level of familiarity with LLMs. Significant differences in the topics males and females are using LLMs for, with males more likely to use them for scientific topics, and females for humanities related topics.

Proposed Solutions: Proactive measures to ensure that both boys and girls develop an equal familiarity with these tools and their limitations from an early age. Addressing this issue is crucial to preventing a widening gender gap in AI-related skills and opportunities. Integrating these systems into their teaching methods. Key areas of focus should include ethics, critical thinking, the limitations of AI, and the environmental costs associated with these technologies.

Environmental Concerns

Environmental issues related to ChatGPT use.

Proposed Solutions: Taking into account environmental issues related to ChatGPT use when considering adopting these tools as part of an educational setting.