The ChatGPT era prompts a boom in A-graded coursework

The core finding

• Some college classes are seeing a boom in students earning A’s, often with the help of AI.
• The concern isn’t a minor bump (A- to A). Per Igor Chirikov (UC Berkeley), who authored a study on AI and grade inflation: “We have a C student who is now an A student” — based on grades from 2018–2025 at a Texas research university.

Why it matters

• Universities were already worried about how many students earn A’s and B’s.
• The new worry: graduates may leave AI-proficient rather than knowledgeable about their actual subjects of study.

The data

• Since ChatGPT’s 2022 release, “excellent” grades rose 30% in classes where AI is useful — e.g. English composition and coding.
• In classes where AI isn’t useful — sculpture, lab-based courses — grades stayed flat.
• The university wasn’t named: a “selective” school with 50,000+ students across all disciplines. Chirikov chose it because its grade data is public and believes the trend is sector-wide, not specific to one school.

The homework-vs-exam pattern

• Excellent grades have risen since the early 2000s, but Chirikov found classes weighting homework over in-class exams show higher grade inflation.
• This suggests unsupervised work is getting an AI-assisted boost.
• Additional pressure: faculty are sometimes incentivized to grade leniently, since student evaluations can factor into promotion.

No silver bullet

• Inflated GPAs aren’t new — students have long picked easier courses for easier A’s. AI “just exacerbates the existing trends.”
• Some professors are already cracking down with handwritten or oral exams (e.g. the return of blue books).

The bottom line

• Chirikov argues for AI-integrated assignments where students may use LLMs but must properly document that use.
• “That’s not an easy process, but we definitely should invest in that more than we do right now.”

Axios, 16 May 2026.