UC Berkeley Professor Used AI to Edit Her Op-Ed Blasting Students' Maths Skills

Key points

• UC Berkeley maths professor Zvezdelina Stankova admitted to using AI to help edit a roughly 2,000-word opinion piece published in the San Francisco Standard.
• The op-ed claimed some of her students arrived at Berkeley up to eight years behind in maths, lacking middle-school knowledge of fractions and basic algebra.
• Stankova blamed UC-system test-blind admissions, which dropped standardised benchmarks like the SAT, for letting underprepared students into the programme.
• The disclosure drew attention because the piece criticised educational standards while relying on an AI tool to polish the argument.

A University of California, Berkeley maths professor published a sharp critique of her students' maths ability last week. Then she admitted she used AI to help edit it.

Zvezdelina Stankova's piece ran in the San Francisco Standard, first flagged by The Guardian. It was about 2,000 words long and made a stark claim: some students in her programme were "five to eight years" behind where they should be, missing foundational maths that most people cover in middle school, things like fractions and basic algebra.

What exactly did she say?

Stankova pointed the finger squarely at test-blind admissions, a policy where universities stop using standardised test scores, like the SAT, as part of the entry decision. The UC system adopted this approach in recent years. Her argument was that without that benchmark, students who were not ready for Berkeley-level maths were still being admitted, and that everyone, students included, was paying a price for it.

It is a legitimate debate worth having. Plenty of educators and researchers disagree about whether the SAT is a fair or useful measure of readiness. But the conversation shifted quickly when Stankova disclosed she had used an AI tool to help edit the piece.

Why does the AI editing disclosure matter?

On its own, using AI to tighten up a draft is not scandalous. Lots of people do it, the same way they might ask a colleague to read something before sending it. The catch here is context. A professor arguing that students lack rigorous skills, in a piece itself polished by an AI tool, is an irony that is hard to ignore.

It also raises a practical question for anyone writing opinion pieces, academic commentary, or anything meant to represent their own thinking: at what point does AI editing shift from light tidying to something more substantial? There is no clear line yet, and most publications do not have formal rules about it.

For students and academics reading this, the takeaway is simple. Disclose AI involvement early, not after the fact. Readers and editors are increasingly expecting it.

What should ordinary readers take from this?

Two things, really. First, the underlying maths-skills debate is real and worth watching. If students are arriving at top universities underprepared, that affects the quality of degrees and, downstream, the workforce. Second, AI editing tools, things like ChatGPT or Grammarly's AI features, are now woven into everyday writing. Free tiers exist on most of them. But using them on anything public, and not saying so, is starting to look like a mistake.

As for privacy: if you paste your writing into an AI tool, that text is typically used to improve the model unless you opt out in settings. Worth checking before you edit anything sensitive.

Common questions

Is it wrong to use AI to edit something you wrote?

Not inherently. Many writers use AI tools the way they would use a spell-checker or a trusted reader. The issue is transparency, especially in journalism or academia, where the expectation is that the voice and reasoning are genuinely yours.

What are test-blind admissions?

Test-blind admissions means a university ignores standardised test scores, like the SAT or ACT, entirely when deciding who gets in. The idea is to reduce barriers for students who do not test well but are otherwise capable. Critics argue it removes a useful academic signal.