2025年学生生成式人工智能调查_12页_599kb

Summary of the 2025 AI Student Survey

Core Content

This 2025 AI Student Survey, conducted by HEPI and Kortext, builds on the 2024 findings to explore how full-time undergraduate students are using AI tools, their attitudes toward them, and their evaluation of institutional responses to AI's growing role in education. The survey included 1,041 students and revealed significant changes in AI adoption and perception over the past year.

Main Findings

AI Usage Among Students

• AI Adoption Has Surged: 92% of students now use AI in some form, up from 66% in 2024.
• GenAI Use for Assessments: 88% of students have used generative AI for assessments, a sharp increase from 53% in 2024.
• Common Uses of AI: The most common uses include explaining concepts (58%), summarising articles (now second most popular), and drafting assessments (25%).
• AI-Generated Text: 18% of students use AI-generated and edited text in their assessments.

Attitudes Toward AI

• Perceived Benefits: Students primarily use AI to save time and improve the quality of their work. Personalised and out-of-hours support are also important.
• Concerns: Many students are hesitant due to fears of academic misconduct accusations and the risk of false or biased results. Environmental impact and data privacy are also concerns, though less prevalent.
• Gender and Socio-Economic Divide: Women are more cautious about AI use and more likely to perceive it as risky. Students from higher socio-economic backgrounds use AI more frequently and for more confident reasons.
• Subject Differences: STEM and Health students are more familiar with and enthusiastic about AI, while Arts and Humanities students are more sceptical.

Institutional Response

• AI Policy Clarity: 80% of students believe their institution has a clear AI policy, up from 66% in 2024.
• Detection Confidence: 76% of students believe their institution can detect AI use in assessments, a significant increase from 65% in 2024.
• Support for AI Skills: Only 36% of students feel they have received support to develop AI skills, despite the majority (67%) believing it is essential.
• Institutional Encouragement: Less than a third of students feel their institution encourages AI use, while nearly a third report that their institution bans it. This highlights a lack of clear guidance.

Key Challenges

• Digital Divide: A persistent divide exists based on gender, socio-economic status, and subject area. This may limit equitable access to AI tools and their benefits.
• Uncertainty About Acceptable Use: While students are aware of AI capabilities, they remain uncertain about what is considered acceptable use, especially in assessments.
• Mixed Institutional Messaging: Some students report confusion due to conflicting messages about AI use, with some institutions discouraging it while others incorporate it into teaching.
• Ethical and Technical Concerns: Students are increasingly aware of issues like bias, hallucinations, and data privacy, but many feel these are not adequately addressed by institutions.

Policy Recommendations

1. Review and Adapt Assessments: Institutions should regularly review their assessment practices and procedures to ensure they are robust against AI use. Assessments that can be easily passed using AI should be redesigned.
2. Enhance Staff AI Literacy: All teaching staff should be trained to understand and use AI tools effectively. Institutions should provide ongoing professional development and support.
3. Develop Nuanced AI Policies: Policies should reflect the inevitability and potential benefits of AI use, while promoting responsible and ethical practices. Institutions should avoid a purely punitive approach.
4. Promote Equitable Access: Institutions should provide AI tools and support to close the digital divide, especially for students from lower socio-economic backgrounds and women.
5. Collaboration and Knowledge Sharing: Institutions should collaborate and share best practices to effectively address the challenges of AI in education. Groups like Universities UK and mission groups should facilitate this.

Conclusion

The 2025 survey underscores the growing integration of AI in higher education, with students using it for a variety of academic purposes. While there are clear benefits, challenges such as the digital divide, ethical concerns, and institutional inconsistency in policies remain. Institutions must adapt to this new reality by supporting students' AI skills, reviewing assessments, and fostering a culture of responsible and equitable AI use.