Major University Study Finds ChatGPT Driving Significant Grade Inflation in AI-Exposed Courses
Key Takeaways
- ▸A grades rose by 13 percentage points (30% relative increase) in courses with heavy AI-exposed writing and coding tasks following ChatGPT's release
- ▸Grade inflation is strongest in courses where homework assignments carry significant weight, indicating AI substitution for student work rather than enhanced learning
- ▸The study analyzed 500,000+ grades across six years (2018-2025) at a major research university using rigorous difference-in-differences methodology
Summary
A comprehensive analysis of over 500,000 grades at a large research university reveals that the introduction of ChatGPT has triggered substantial grade inflation in courses heavily reliant on AI-exposed tasks like writing and coding. The study, conducted by Igor Chirikov and published May 13, 2026, tracked grade distributions from 2018 to 2025 using a difference-in-differences methodology to isolate ChatGPT's impact post-release in November 2022.
The research found a striking 13 percentage point increase in the share of A grades in affected courses—representing approximately 30% growth relative to 2022 baseline levels. Grade inflation was most pronounced in courses where homework assignments carried greater weight in final grade calculations, suggesting that students are using generative AI tools to substitute for actual work rather than leveraging them as genuine learning aids.
The findings raise critical questions about grade validity and the real-world preparation of students in the ChatGPT era. While institutions continue grappling with AI integration policies, this data suggests that without proper safeguards and assessment redesign, grades may no longer reliably signal student competency in foundational skills like writing and coding—competencies essential for workforce readiness.
- Universities may need to redesign assessment methods and establish clearer AI usage policies to maintain the informational value of grades
Editorial Opinion
This research provides sobering empirical evidence that ChatGPT is fundamentally disrupting the integrity of academic assessment, at least in its current form. The 30% jump in top grades in just a few years suggests widespread adoption of generative AI for assignment completion rather than supplementary learning—a troubling signal for employers and graduate programs relying on grades as quality signals. While AI tools have legitimate educational value, institutions need urgent action on assessment redesign and transparent AI usage policies before grades become merely decorative credentials.
