Someone I know will be applying to universities in the US for their PhD. To that end, he sought LORs from his undergrad instructors. However, they told him to write those himself, and to send those to them.
He used ChatGPT to generate the LORs. He didn't fabricate anything though. He just used ChatGPT to write truthful things about himself in the recommendation letter. Is that okay?
Sure, assuming that your friend entered truthful information I guess it really doesn't matter. I don't think there is anything unethical about it (certainly no more unethical than writing your own letter - which is pretty commonly accepted).
I would not do it only because I don't think the results are very good. But I wouldn't worry about the ethics exactly. This isn't a school assignment. Personally, I think that ChatGPT has an awkward sort of stilted writing style. It sounds very...generic. I guess it's probably an improvement over some people's writing but it lacks character. It is a good free-ish alternative to using the AI assisted editing tools available though (I think Word has one built in these days).
Aside from the question of whether this is ethical or not, it's a bad idea because it will not produce a persuasive letter in the professor's style, and all of the letters will end up sounding about the same. Letters of recommendation need to be highly individual and personalized---they need to say what the WRITER knows about the applicant from direct personal interactions with and observations of the person. They should not just mechanically parrot what is already obvious in the applicant's CV and teaching and research statements. BAD: "Jane Doe has published six papers in prestigious journals and presented seven conferences in three countries." That's straight out of the CV. GOOD: "Jane Doe was a research assistant on my Project X. We strategized the research design together, and her input resulted in several key improvements to the hypotheses and methods. She oversaw three undergraduate researchers, performed the qualitative data analysis of the interview results, and wrote the literature review. I especially noted that she did not lose heart when the first part of the study did not yield significant results, but took this in stride as a normal part of scientific research." AI can write something more elegant than that, but not more personalized. This personalization is at the heart of all letters of recommendation. An AI-generated letter is (a) likely to be detected and (b) unlikely to impress---two big risks. I wouldn't do it. It conveys that the supposed letter writer doesn't feel strongly enough about the candidate to put in the effort and write an original letter. The same goes for professors who ask students to write their own letters. It also says that the professor doesn't care about the quality of the letters they put their signature to, harming the professor's credibility and hence the applicant. Finally, AI is not in a position to judge the quality of the applicant's work. I see no advantages at all to an AI-generated letter, given the real purpose of these letters, which is to tell the admissions committee about the writer's PERSONAL knowledge, experiences, perceptions, feelings, and professional judgments related to the candidate.
I personally would not ask any BOT to write letters of recommendation for my students.
These have to be personal and highlight unique qualities of a particular student. Also I like to include anecdotes about my interaction with a particular students. The letter is then less bland and helps distinguish that student from the mass of other students applying for the same position.
I personally think this is “ok” in the moral and ethical sense, but be aware that a lot of faculty, who should know better, are really afraid of ChatGPT, don’t know its capabilities, and imagine that it is somehow spitting out Pulitzer Prize worthy tomes with the simplest of prompts and therefor using it is always dishonest and immoral.
I’m an enthusiastic user of ChatGPT for all kinds of things, like generating outlines, cleaning up email communications, getting cocktail recipes and travel itineraries, and getting a decent workout routine. If your friend spends time refining their prompts, using ChatGPT to clean up and edit, and then personalizing the ChatGPt output, this can be quite effective.
I’m seriously bothered, however, by the fact that the faculty members told your friend to do this. At this stage, a faculty member can write a much better letter about an undergrad than that undergrad can write about themselves. This is terribly irresponsible of them.
Generally it is quite clearly stated by graduate schools that no part of letter should be written by the student and it is duty of the student to let the recommendor know about this policy.
ChatGPT is merely a tool which can be used to refine the writing of someone's rough draft so that's not the problem. I see the problem of instructor making the student read it.
Often it is easy for graduate schools to tell that the letter is not written by instructor because of kind of way it talks about student which chatgpt might be able to parse through.
I think it is as unethical as writing your own LOR, because at the end of the day it contains information that you know about yourself, and not what your professor knows. Ofcourse, you can always talk to the professor about yourself and discuss what to write, but they will include it from their perspective not yours, and that is the important thing.
I personally think it is very lazy and wrong on your friend's professor's part to ask the student to write a letter which they will just sign. They can ask for a CV, or an SOP, or a personal statement, or any other info which will help them to write an LOR, but not ask the student to write it directly. And if they don't have the time to write it themselves, they should politely refuse to provide it.
