ChatGPT Is Flooding College Transcripts With A's — and Employers Are Noticing

A new study from UC Berkeley has quantified what many professors have suspected: AI tools like ChatGPT are directly inflating college grades. Analyzing more than half a million grades from 2018 to 2025 at a large public Texas university, researcher Igor Chirikov found that courses heavy in writing and coding — the tasks most susceptible to AI assistance — saw roughly 30% more A's awarded after ChatGPT's launch in late 2022. Crucially, the two groups of courses tracked nearly identically before that date, isolating AI as the likely driver. The inflation is especially pronounced in take-home assignments, where AI access is unrestricted.

Chirikov's core concern isn't just grade fairness — it's what he calls the erosion of 'productive struggle,' the cognitive friction that actually builds skills. If students are outsourcing writing and coding to AI models, they may be producing polished-looking work while developing fewer underlying competencies. An A grade, he now argues, may signal access to a better AI model or skill at prompting it — not mastery of the subject matter.

The timing is particularly damaging for the job market. After years of deemphasizing GPA, employers are leaning back on it as a hiring filter amid a flood of entry-level applicants. The share of employers using GPA in hiring decisions has climbed from 37% in 2023 to 42% today, per the National Association of Colleges and Employers. Barclays and Morgan Stanley maintain explicit GPA minimums for some internship roles. On Handshake, the entry-level jobs platform, nearly a quarter of GPA-requiring postings now demand a 3.5 or higher — up from just 9% in 2020. The cruel irony: the metric employers are increasingly relying on is being quietly hollowed out.

Elite universities are beginning to sound the alarm. Harvard is holding a faculty vote this week on a proposal to cap the number of A's given out, after a February report warned that current grading policies prevent meaningful comparison of student performance. Yale's April report was even more blunt, stating flatly that grades at Yale — like at many peer institutions — no longer communicate what students have actually learned. Meanwhile, educators are adapting in real time: a Wharton professor now weights in-class exams and quizzes more heavily after realizing AI could score a perfect 100 on her standard homework assignments.