University Study Reveals AI Bans Harm Law Students' Performance
A two-year university study led by legal scholar Schrepel found that banning AI from classrooms leaves students worse off, with even untrained AI use leading to better outcomes than outright prohibition. The controlled experiment, involving law students revising a provision of the EU AI Act, demonstrated that students without AI consistently performed the worst. For broader context, explore our AI News.
The Classroom Experiment: AI Access vs. Prohibition
The study, conducted in 2024 and repeated in 2025, involved law students tasked with a complex legal revision. Participants were divided into three distinct groups to assess the impact of AI integration:
- No-AI Group: Students in this group were explicitly forbidden from using ChatGPT or similar AI tools.
- Unguided AI Group: These students were allowed to use ChatGPT for suggestions but received no formal training on how to use the tool effectively.
- Trained AI Group: This cohort received hands-on instruction and guidance on how to best utilize AI for their task.
The initial experiment in 2024 included 66 students, expanding to 164 participants in 2025, providing a robust dataset for analysis.
Key Findings: The Detriment of AI Bans
The results consistently showed that the group without AI access performed the worst. In both years of the study, the no-AI group frequently struggled, often running out of ideas within ten to fifteen minutes. This suggests that a complete ban on AI tools can significantly limit students' capacity for sustained ideation and problem-solving in complex tasks.
Interestingly, the study also revealed that every subgroup using ChatGPT, even without formal training, retained at least one misleading or legally extraneous term from the AI's output. However, despite these imperfections, these students still outperformed their AI-free counterparts. This highlights a crucial point: while AI outputs require critical evaluation, the mere exposure and use of these tools can foster a natural learning process where students begin to identify AI's weaknesses.
Evolving Impact of Training and Familiarity
A notable shift occurred between the two years of the experiment regarding the impact of formal training. In 2024, the trained AI group showed a clear advantage over the unguided AI group. However, by 2025, this performance gap had nearly vanished. Schrepel, the study's author, initially assumed that AI's benefits were contingent on structured training, but the evolving results challenged this premise.
This diminishing gap implies that as students gain more everyday familiarity with AI chatbots, the need for extensive formal instruction may decrease. Students are, through regular interaction, developing an intuitive understanding of how to work with and critically assess AI-generated content.
Why This Matters for Education and Policy
The findings from this two-year study carry significant implications for educational institutions and policymakers. The study strongly suggests that blanket AI bans produce measurably worse average outcomes than allowing AI use. Instead of prohibition, universities should consider policies that embrace AI, focusing on teaching students how to critically engage with these tools rather than isolating them from a technology that is rapidly becoming ubiquitous.
For educators, this means shifting from a mindset of restriction to one of integration and critical literacy. Equipping students with the skills to effectively use and evaluate AI tools will be crucial for their future success in a world increasingly shaped by artificial intelligence.
Conclusion: Embracing AI with Critical Engagement
The UC Berkeley Law study provides compelling evidence that outright bans on AI in education are counterproductive. While AI tools like ChatGPT are not flawless and require careful scrutiny, their integration into learning environments, even without extensive formal training, can lead to better student outcomes than their complete exclusion. As AI continues to advance, fostering an environment where students learn to critically engage with these powerful tools will be paramount for preparing them for the challenges and opportunities of the future.
Sources
Recommended AI tools
Studocu
Productivity & Collaboration
Empower everyone to excel at their studies
Knowt
Productivity & Collaboration
#1 Free Quizlet Alternative with AI Study Tools
DataCamp
Data Analytics
Master Data & AI Skills. Learn by Doing.
Cognito
Conversational AI
Instant feedback for smarter learning
ReadTheory
Productivity & Collaboration
Adaptive reading practice for measurable growth
Up Learn
Conversational AI
A*/A guaranteed or your money back
About the Author
Albert Schaper is a co-founder of Best-AI.org. He focuses on product strategy, AI adoption, practical tool selection, and educational content that helps users compare AI products with clearer context.
More from AlbertWas this article helpful?
Found outdated info or have suggestions? Send us a note.