FEATURED
Key Takeaways
- Princeton University plans to reintroduce proctors in exam rooms, revising its 133-year-old Honor Code to combat increasing concerns over AI-related cheating.
- The decision comes after numerous reports from students and faculty about the perceived rise in cheating, facilitated by easy access to generative AI.
- Critics, including an AI strategist, suggest that proctoring may not be a sustainable solution and recommend clearer rules and teaching students responsible AI use instead.
Princeton University recently decided to adopt a proposal that would reintroduce proctors to every exam room over AI cheating fears.
The proposal would revise Princeton’s 133-year-old Honor Code, an agreement between students and faculty established in 1893. Previously, students pledged not to take unfair advantage during examinations. In exchange, faculty proctors were not present in the exam rooms.
The proposal cites “the advent of generative artificial intelligence products” as one of the primary concerns which prompted its development. As a result, instructors will “act as additional observers” in the exam room.
“Over the past few years, and with increasing frequency over the last six months, significant numbers of undergraduate students and faculty members have requested that the Office of the Dean of the College consider a policy change toward proctored examinations, given their perception that cheating on in-class exams has become widespread,” it states.
The proposal also states that staff and students are concerned the ease of access to generative AI on small devices has “changed the external appearance of misconduct” and lowered “the barrier to gaining unfair advantage.”
In addition, many reports to the Honor Committee are now submitted anonymously due to students’ fear of doxxing or social media shaming. This reluctance to report openly has made it harder for the school to investigate and follow up on potential violations.
Reached for comment, an AI and digital strategist told The College Fix the proposal may not be the best long-term solution.
“I understand why Princeton is responding to concerns about academic dishonesty,” Stephen Rowe said, “but I would be careful about treating proctoring as the long-term answer.”
“This is just the beginning of a new learning paradigm for students,” he said.
Rowe said the benefits and drawbacks of proctoring exams depend on the type of information students are taught and the way faculty expect students to learn the material.
“If the goal is research, argument, analysis, writing, technical skill, or professional judgment, students should often be expected to use AI responsibly and explain how they used it,” he said.
He also commented on the “uneven” nature of AI use in academic settings, saying, “Some students use it as a shortcut. Others use it as a tutor, writing coach, research assistant, brainstorming partner, coding helper, or debate partner.”
“That range is exactly why schools need clearer rules,” he said.
Rowe recommended that educators focus on “assignments tied to specific course materials or original student reasoning” like “oral defense, drafts, [and] source evaluation.”
He also suggested that colleges “teach students how to use AI with integrity, through clear policies, discipline-specific rules, and assignments that separate prohibited substitution from permitted assistance.”
Rowe’s call for caution in the adoption of widespread proctoring echoes Dartmouth’s 2024 response to AI, which gave instructors more discretion over “how they administer examinations.” It allowed proctoring as one option.
Reached for comment, The Alliance for Secure AI, a nonprofit that aims to “defend humanity” in the age of AI, said current federal policy is insufficient.
“Princeton University’s actions highlight how current policy frameworks are not enough in the age of AI,” spokesperson Peyton Hornberger said.
“With no national standard in place, universities are being forced to take individual measures,” she said.
Princeton’s online guidelines on AI use encourage students to take responsibility for AI outputs, protect privacy, verify accuracy, watch for bias, remain transparent, respect copyright, and uphold ethical standards. The guidelines warn against over-reliance without verification and using tools without understanding their limitations.
“These principles are admirable,” Hornberger said. “However, they may place a cognitive burden on students.”
She added that, “AI can be used more securely at universities by putting students first,” which “means creating more comprehensive AI literacy education and giving them the opportunity to think critically both with and without the technology.”
“Students and faculty alike are adjusting to the AI era, and everyone must have the space to learn,” she said.
The Fix reached out to Princeton University media relations and the dean’s office for more information about the new policy, but did not receive a response.
