ChatGPT Use Linked to Grade Inflation in Texas University Courses
A UC Berkeley study of over 500,000 grades at a Texas university found that courses relying on writing and coding saw approximately 30% more A grades after ChatGPT's launch, while students' actual learning may not have increased. As employers increasingly use GPA for hiring, the inflation of grades via AI assistance is making academic performance a less reliable measure of candidate capability.
What happened
A study by Igor Chirikov at UC Berkeley's Center for Studies in Higher Education analysed over 500,000 grades from 2018 to 2025 at a large public university in Texas. The study found that professors in courses dependent on writing assignments and coding tasksโareas exposed to AI useโgave out about 30% more A grades and fewer A-minus and B-plus grades after ChatGPT's launch in late 2022. The grade inflation effect was particularly pronounced in courses with more take-home assignments where AI tools can more easily assist students.
Reported impact
The study indicates that students have relied on generative AI to achieve better grades, not that they have learned more. Employers are increasingly concerned that AI-boosted grades no longer reliably indicate student capability or learning. As of 2026, 42% of companies consider GPA in hiring decisionsโup from 37% in 2023โand firms including Barclays and Morgan Stanley use GPA minimums for some internship roles, making the reliability of grade-based assessment more critical.
Index analysis
Grade inflation has been a long-standing issue on college campuses, but AI tools have made it harder to assess the quality of students' work. The Berkeley study compared AI-exposed classes with less AI-dependent ones and found little difference in grade patterns through 2022; divergence occurred afterward. Learning requires productive struggle that is eroded by heavy AI use, raising questions about whether graduates develop critical cognitive skills.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.