A mathematics professor from the University of California, Berkeley, has sparked discussion over significant gaps in students' mathematical knowledge in a recent op-ed for the San Francisco Standard. In the essay, Zvezdelina Stankova expressed concerns that some of her students are lagging behind by "five to eight years," lacking foundational skills in fractions and basic algebra. She attributed this issue to the UC system's test-blind admissions policy, suggesting that the removal of standardized tests like the SAT has allowed underprepared students to enter Berkeley’s rigorous mathematics program.
The op-ed, which spanned 2,000 words, caught the attention of journalists from the Daily Californian, the student newspaper at Berkeley. They observed that the language in the piece felt reminiscent of artificial intelligence writing. In a follow-up investigation, sophomore Francis Luo employed AI-detection software called Pangram, which flagged that around 33% of the text might be generated or assisted by AI.
In light of this finding, Stankova acknowledged her use of AI technology for editing purposes. She clarified that while she utilized AI to identify relevant documents and literature, the final article represented the cumulative effort of numerous hours of human work, including about 80 hours of her own investment.
When queried about the ethical implications of AI use in their publication, the Standard stated that all articles should be human-authored and under human accountability. They emphasized that their collaboration with Stankova was based on the understanding that her piece reflected her and her colleagues' original insights and analysis.
The revelation about AI assistance triggered a broader dialogue on social media regarding its role in academia. Critics labeled the situation as ironic, considering the op-ed's emphasis on students' lack of preparedness, arguing for the need for wholly original content. On the other hand, some supporters contended that the AI's involvement was trivial and diverged focus from the core argument.
Stankova responded via email, stating that the concerns about AI usage were peripheral and diverted attention from the extensive effort her team has put into advocating for the reinstatement of standardized testing in admissions.
Discussions about whether to reinstate tests like the SAT or ACT in the UC admissions process have intensified in the last few years. The University of California’s governing body voted in 2020 to eliminate these tests as a requirement, a move echoed by institutions like Harvard, MIT, and Yale, which later reinstated testing requirements.
In her op-ed, Stankova provided data highlighting a "severe deficit" in student preparation for calculus courses, a situation that reportedly worsened following the shift to test-blind admissions. Before publishing, she was among more than 3,000 UC faculty members who signed a letter in June advocating for the return of standardized testing, mentioning that AI editing was also employed in that context.
In late July, the UC academic senate announced plans to review the potential reinstatement of standardized tests for admissions. Any changes would impact admissions for fall 2028 at the earliest. The UC system is also home to an "AI council" dedicated to fostering AI initiatives and resources. Camille Crittenden, a council member, indicated uncertainty regarding guidelines for faculty on AI usage in personal research or editorials, while noting that many AI detection tools, including Pangram, lack precision and nuance.
Pangram claims to accurately identify AI-generated content 99.66% of the time, reportedly generating one false positive for every 24,000 documents analyzed.
