Universities Face Grade Inflation Crisis as Students Use AI to Bypass Learning
A working paper from UC Berkeley researchers found that courses relying on take-home assignments and essays saw a 30 percent jump in "A" grades since ChatGPT entered widespread use. The study analyzed over 500,000 student enrollments across 84 departments at a large Texas university from 2018 to 2025.
The research identifies three ways students use generative AI in their coursework. Augmentation means using AI as a research tool while the student does most of the work. Reinstatement involves creating new AI-based tasks. Displacement is when AI completes the work entirely-writing essays or solving coding problems without student input.
Only augmentation and reinstatement correlate with actual learning. Displacement improves grades while eroding skill development.
Where AI Cheating Thrives
Students exploit AI most in unsupervised settings: take-home assignments, essays, and homework. Proctored exams, oral presentations, and in-class discussions remain harder targets for AI automation.
The concentration of grade increases in writing and coding courses suggests students are using AI to complete assignments rather than learn from them. As job competition intensifies and AI adoption accelerates across industries, students have clear incentive to chase higher GPAs through any available means.
The Workforce Problem
Grade inflation masks a deeper issue: students may graduate without core competencies in the very areas where AI excels. When AI displaces skill-building during education, graduates enter the workforce unprepared for foundational tasks.
This creates a feedback loop. Employers hire workers dependent on AI. Industries accelerate automation because staff lack the skills to perform jobs manually. Job displacement accelerates.
Employers also face a practical problem: inflated grades make it harder to identify genuinely strong candidates from weaker ones during hiring.
Universities Begin Responding
Some institutions are taking action, though effectiveness remains unclear. Princeton's faculty voted this week to overturn a 133-year-old honor code that allowed students to take in-person exams without faculty supervision. The move followed a survey showing roughly 30 percent of seniors admitted to cheating, mostly through generative AI.
Harvard faculty are voting on a proposal to cap "A" grades at no more than 20 percent of each class.
These measures address symptoms rather than root causes. Four years into widespread AI for Education adoption, most American universities still lack coherent policies on AI use in coursework.
Your membership also unlocks:
