Should faculty use AI to write recommendation letters?


If you attended an academic conference in the past six months, was the keynote address about AI? Likely so, and for good reason! It has been approximately a year since the advent of ChatGPT, and we are still trying to adapt.

The Instructional Technology Council recently held a very popular summit on the topic in November. We are likely to see a continuation of professional development that incorporates AI content, and as it expands, more questions will come.

In one of the keynotes that I heard this fall, the speaker championed the use of generative AI for writing recommendation letters. Once my heart restarted, I looked around the room to check reactions, but my colleagues seemed relaxed and frustratingly calmer than I was. Was this already happening? Was it ethical?

The human touch

To provide context to my reaction, I am a communication professor for a community college in North Carolina. Although I’ve taught classes like public speaking and mass communication, I primarily teach interpersonal communication or, in common terms, a class on, “how to understand and improve significant relationships.”

This article is part of a biweekly series provided by the Instructional Technology Council, an affiliated council of the American Association of Community Colleges.

Almost all of us would say that significant relationships in our lives have been improved in some way by letters. Reflecting on this, I remembered that my parents began their relationship through letters, and I still treasure letters exchanged with friends in college. Letters aren’t just words; they are memories. Whether it is a note of encouragement, an expression of sympathy, a commendation letter, or a note in a child’s lunchbox, the practice of carefully curating words in the mind and cultivating them in the imagination for an audience of one is distinctly human.

I recoil at the thought of using AI to author any personal or professional letter. The idea seems like the scene from “Ferris Bueller’s Day Off” when Ferris creates an elaborate mechanism for skipping school to fake his presence (and illness) in his bedroom whenever someone knocks on his door. It was clever, no doubt, but deceptive.

Pondering two points

As I pondered this, as well as the counterpoints my colleagues at the conference might’ve been feeling, I realized two important things. First, letter-writing is sometimes an onerous task. As Professor James H. Sweet wrote about this subject in 2022, “Our profession is dependent on the goodwill of others, but letter writing is time-consuming and frustrating… I was devoting far too much time and energy to letters that seemed superfluous or unnecessary.”

Professors do wonder, “How often are these letters read?” and “Do they matter?” We are often unsure. While I still have a recommendation letter from my favorite teacher in my files, I could not locate many others if needed.

Secondly, I considered that when a student asks for a recommendation, I almost always say “yes,” even in instances when I haven’t had many interactions with the student from which to draw observations and conclusions. The letter that results in those circumstances may not be deceptive per se, but it is often general in wording featuring a few recycled phrases. (Isn’t that description eerily similar to the writing of a generative AI program?)

Giving it a try

Maybe I could only understand this question by trying a program. I visited to see what AI could do. I entered the type of letter I was writing (student recommendation) and entered three characteristics of a hypothetical student in a field box that I wished to highlight in the letter (strong work ethic, collaborative attitude and resilience). With one click, the AI delivered a letter to me that wasn’t half bad. It was half good.

Like many technical conundrums, the answer would not be as clear — black and white or right or wrong — as one would like. I could understand why a faculty member might use this tool as a starting point for a general letter of rec.

Even history’s most celebrated writers don’t agree on the quality and endurance of letters. Goethe wrote, “Letters are among the most significant memorial a person can leave behind them,” while Thoreau said, “I have received no more than one or two letters in my life that were worth the postage.”

If you find yourself more on the side of keeping letter writing as human as possible, consider that no machine understands the obstacles students face in our classrooms or the resilience they show. We know our students’ unique situations and the context of their journey.

Additionally, no generative AI program will ever convey the warmth and specificity of a teacher who deeply cares. We are still the best “creatures” to personally recommend others, whether we use the most advanced tools at our disposal or not. Maybe Hemingway said it best: “Or don’t you like to write letters? I do because it’s such a swell way to keep from working and yet feel you’ve done something.”

About the Author

Brittany Hochstaetter
Brittany Hochstaetter is a professor of communication at Wake Tech Community College in Raleigh, North Carolina. She serves on the board of directors of the Instructional Technology Council.