AI Just Destroyed College Degrees
By Shizza Farooqui
For decades, universities sold students the same promise:
study hard, get good grades, earn a respected degree, and employers will reward you for it.
Artificial intelligence may now be breaking that entire system at once.
The Grade Inflation Crisis Nobody Warned You About
A major new study from the University of California, Berkeley analyzed more than 500,000 student-course enrollments across 84 departments at a large Texas university between 2018 and 2025. The finding was simple but alarming: after ChatGPT became mainstream, classes heavy in writing and coding began producing dramatically more A grades.
Courses with stronger AI exposure saw A grades rise by roughly 13 percentage points, or around 30% relative to pre-ChatGPT levels.
The most important detail is that the gains were concentrated at the top. AI was not primarily lifting struggling students. It was converting A-minus and B-plus students into straight-A students.
The result is not just grade inflation.
It is grade compression.
Harvard No Longer Trusts Its Own Transcripts
At Harvard, the median graduation GPA reportedly sat at 3.67 for years before climbing toward a flat 4.00. The Sophia Freund Prize, awarded to the university’s top graduating student, went to just two students in 2010-11. By 2024-25, it went to 55 students.
Not because intelligence suddenly exploded.
Because the metric itself stopped working.
Faculty members reportedly described grading as “a race to the bottom,” while honors committees increasingly relied on private conversations to identify “true stars” hidden inside inflated transcripts.
Harvard is now replacing GPA internally with Average Percentile Rank, or APR, for honors and fellowship decisions. The system will not appear on transcripts because universities themselves increasingly understand that transcripts may no longer reflect actual mastery.
The deeper fear emerging inside academia is not simply cheating.
It is what some researchers now call “cognitive surrender.”
Students are outsourcing parts of the learning process itself.
Research, writing, coding, editing, summarizing and brainstorming can now all be assisted by AI systems capable of producing polished work in seconds. Universities are struggling to determine whether submitted work still represents genuine understanding.
Students Are Deliberately Writing Worse. Here Is Why.
Some institutions are responding aggressively.
Princeton University recently voted to fully proctor in-person exams, fundamentally reshaping parts of its 133-year-old honor system after AI-related cheating concerns intensified. Surveys found large numbers of graduating students admitted using AI tools even in assignments where they were prohibited.
At the same time, AI detection systems themselves remain controversial because they frequently produce false positives, especially against non-native English speakers.
That has created another bizarre side effect: some students now intentionally make their writing worse to avoid being falsely accused of using AI.
Only 2% of Employers Still Trust Resumes
But the crisis extends far beyond campuses.
Employers are already losing trust in traditional hiring signals. A new Criteria Corp and Lighthouse Research survey of nearly 1,000 hiring leaders found that only one-third of employers remain highly confident that resumes accurately reflect candidates’ real abilities. Ninety-two percent said AI-generated resumes are now commonplace. Only 2% of employers said resumes remain their most trusted hiring signal.
Many companies are now shifting toward structured interviews, simulations, live testing and direct skills assessments instead of relying on GPA or written applications.
AI Was Supposed to Level the Playing Field
And there is another uncomfortable layer beneath all of this: the AI advantage itself increasingly reflects wealth inequality.
Students from wealthier families can afford stronger premium AI tools with fewer usage restrictions, while lower-income students often use AI less frequently or with weaker free versions. Researchers warn this could create a two-tier educational system where the students already most advantaged become even more competitive through AI-assisted performance.
The irony is brutal.
AI was supposed to democratize opportunity.
Instead, it may be accelerating the collapse of trust in the very systems used to measure human ability.
And as universities, employers and students all begin questioning whether grades, degrees and resumes still mean anything at all, one uncomfortable possibility is starting to emerge:
the age of measurable human intelligence may be ending.
By Shizza Farooqui
Sources
Wall Street Journal, UC Berkeley, Harvard University, Criteria Corp, Lighthouse Research, Princeton University, Gallup, HEPI, Harvard Business School, Dallas Express, DeepLearning.ai, arXiv
