Growing use of generative AI by students is leading to rising grade inflation at colleges, according to a working paper released this week by the University of California, Berkeley.
There are three ways in which generative AI can be used by students: augmentation, where the tools play a supporting role in helping with things like research while the student does most of the work themselves; reinstatement of new AI-based tasks; or by displacement, where it completely automates work that a student would otherwise do themselves, such as writing an essay. All three use cases can improve grades, while only augmentation and reinstatement can correlate more with actual learning and skill development.
Certain academic tasks, such as unsupervised take-home assignments, term papers, and other assignments, are perfect opportunities for AI replacement, as opposed to proctored exams, oral presentations, or class discussions.
As part of the study, Igor Chirikov, a senior researcher at UC Berkeley, analyzed more than 500,000 course registrations across 84 departments at a large Texas university between 2018 and 2025. He found that grade increases were primarily concentrated in courses “with a higher share of writing and coding tasks” where homework assignments carried the most weight, concluding that students use AI to cheat on certain assignments and get better grades. Overall, researchers found that “AI-exposed courses” saw a 30% increase in “A” grades since ChatGPT hit the market.
This isn’t particularly shocking; This is a generative AI use case as old as the dawn of ChatGPT. Additionally, a student’s GPA could be decisive for their future, determining their acceptance into postgraduate academic programs and lucrative early career job opportunities. So, in a world where most industries are moving towards AI, often to the detriment of the recent graduate job market, it makes sense that the average student is looking for an easy way to secure their future.
What’s interesting is that, four years after the widespread presence of generative AI in our daily lives, the study shows that American universities have not yet caught up with its consequences.
With increased grade inflation driven by AI, employers will have a harder time weeding out the strongest young graduate applicants, the study found. But more importantly, this increased reliance on AI in academia is sure to create an incompetent, AI-dependent workforce.
“If AI replaces skill-building tasks during learning, students could graduate with weaker abilities precisely in areas where AI is strongest, thereby reinforcing a feedback loop between AI in education and AI in production that could accelerate automation,” writes Chirikov.
Thus, a university system that responds to grade inflation with AI would create a workforce that does not know how to perform the core tasks of their jobs, which in turn would create increased reliance on AI in the workforce and even more massive automation of jobs, on the path to a much-feared armageddon of AI employment that some experts say is already underway in some sectors.
Some universities are considering taking action against this grade inflation, but whether the planned measures will actually be effective remains up for debate. At Princeton, where about 30% of seniors admitted to cheating primarily through generative AI in a recent survey, professors voted this week to rescind a 133-year-old honor code that allowed students to take exams in person without supervision from a faculty member.
Meanwhile, at Harvard, professors are voting on a proposal to limit A grades to no more than 20 percent of the class.