Princeton’s Return to Proctored Exams Reflects Changing Times

The following is a fable in the form of a dialogue. No large language model was used in creating it. 

“‘The Economic Influence of the Developments in Shipbuilding Techniques, 1450 to 1485.’ Is this your own work, Mr. Dixon?” 

“Yes, Professor.” 

“You wrote it yourself? All 5,000 or so words?” 

“Yes.” 

“I’m sure you know the university’s rules. You’re allowed to use the internet in your research, but if you quote sources you have to credit them. Even if you paraphrase them, as it were, you have to say where you found what you’re paraphrasing and who wrote the original version. You’re aware of that, aren’t you?”

“Yes, of course.” 

“And you must never use artificial intelligence to do your writing for you. The writing has to be your own, not Claude’s or Gemini’s or some other chatbot’s. Did I mention that? You aren’t permitted to use AI to make a first draft, then go over it and just change a word here and there.” 

“I used a spell checker. That’s acceptable, isn’t it?” 

“Yes. So is a grammar checker. But a paragraph generator isn’t. Is that understood, Mr. Dixon?” 

“Yes, Professor.” 

“Then we can get down to business. I’ve marked 10 sentences in your paper. Each of them is about something you probably didn’t know before writing the paper. I’ll read them to you and ask you questions about them. Like, ‘what does this mean?’ or ‘what country is this person from’ or ‘what trade did this company engage in?’ Is that clear?” 

“Yes, Professor.” 

“Since you’re thoroughly familiar with the sources you cite, you should have no trouble answering the questions without consulting your paper or your notes. Let’s see. Is there something else? Oh yes. I’m going to have to give you a grade based on your answers. Ten right is an A, nine right is a B, eight is a C, seven is a D, and anything below that is an F. Is that clear?” 

“Yes.” 

“Splendid! Then let’s begin. First question: You wrote ‘Of Columbus’ three ships, the Niña and Pinta were caravels; the Santa Maria was a carrack.’ What’s the difference between a caravel and a carrack?” 

The moral of the fable could be that this is one way to uncover AI cheating. Or it could be saying that personal interaction between docent and disciple is the best way to teach effectively. Some background: The title of Dixon’s paper is taken verbatim from Kingsley Amis’ first and most famous novel, Lucky Jim, published in 1954. Jim Dixon, the central character, is a harassed junior lecturer at a redbrick university, not a student as in this fable. His boss, Professor Welch, is responsible for much of the harassment. As portrayed by Amis, Dixon’s adherence to academic norms is not overly devout. If he had had access to AI, he might have used it in writing the paper and consequently have been unable to answer Welch’s question. But in view of his unscholarliness, he might have been unable to answer the question even if he never used AI. In the novel, in a twist that would be doubly ironic in the Age of AI, Dixon himself falls victim to a plagiarist. 

Amis was a visiting fellow in creative writing at Princeton in 1958-59. His 1963 campus novel, One Fat Englishman, is set in “Budweiser College,” modeled on Princeton. 

I read with great interest and sadness about Princeton’s decision to begin proctoring exams (July/August issue). I and my classmates regarded the Honor Code with great pride; not all schools trusted their students to abide by such a code. I still have vivid memories of the solemn Honor Code session in Alexander Hall, and the signing of the Honor Code on all my papers, including my thesis.

I have been teaching first-year medial students at Harvard Medical School for more than 30 years. Part of their education is focused on “professionalism.” Much of the teaching is about “integrity.” The definition of integrity I use in these discussions is: “Integrity is about what you do when no one is looking.” I tell the students that if I ask them about a lab value for a patient they were supposed to check and they make up a number because they forgot to do it or woke up late that day, they can’t be on my team. They can tell me they don’t know because they overslept and didn’t have time, but they can’t make up a number; the absence of integrity disqualifies them from being a doctor and they can’t be on my team.

There is much discussion today about the pressures on college students and the competition for graduate and professional school admissions and top jobs with financial services firms. Whatever grades you have, if you don’t have integrity, nothing else matters — you can’t be on my team.

With respect to the shocking article in the July/August edition of PAW announcing the return to proctored exams: Do you suppose that it has ever occurred to anyone in the administration that Princeton has been admitting the wrong people?

I can state with a certainty that I never “gave nor received assistance” or cheated in any other way while at Princeton. However, this does not mean that I have not witnessed others doing it. And, no, I would not report them. In any event, cheating had a limited impact back then, especially since exams were often open book and/or open notes.

However, times have changed. My wife is a teacher at a German school, and while cell phones are banned during tests, people manage to sneak in cell phones, or have someone in the bathroom with those same phones. And quite often a student whose English is atrocious comes up with a composition which is nothing less than perfect. There are sites which detect AI use, but that evidence is not admissible in a German school. The only recourse a teacher has is to ask the student to come for an oral examination.

Thus, there needs to be a way to make sure those things do not happen, or at least make it much more difficult to do so. And, as much as I hate the idea of proctors at a Princeton exam, I see that there now might be a need for it — and also the need for a way to deal with AI use if same has been detected.

While I agree with the importance of actual adherence to the Honor Code by students, the pearl clutching over having proctors at exams is a bit much. The fact is that Princeton has not trusted its students to be honest since at least the ’70s in the situations where it matters. When I was a freshman in 1974, my biology final conflicted with my sister’s college graduation. I requested to take the exam early and immediately leave for the summer so I could attend the graduation, but was told that Princeton’s exam policy prohibited that, notwithstanding the Honor Code. Whether proctors are present or absent in exams rooms is of no benefit to students; whether students are forced to miss important family events because the University does not believe its students are honest is far more serious.

The Class of 2030 has a unique opportunity — restoring honor to Princeton. By their conduct, they can convince the administration that Princeton students can again be trusted with proctor-less exams.

I have faith in this class. I don’t think they will make excuses. They know that AI is only the latest in cheating enabling technologies that have tempted Princetonians — just like photocopying machines, computers, the internet, and smart phones. They know that Princetonians observed the Honor Code during the tough job market of the Depression.

When almost a third of a class self-report cheating, it becomes clear that honesty has become a lifestyle choice rather than a core value. “Blaming it on AI” deflects from the need to address a far deeper problem.

Class of 2030: I hope you take up this challenge and by your graduation, I can read about the restoration of the original Honor Code.