13

I am in the job market this year for tenure-track positions.

I was wondering if you could tell me how good/bad it is to create cover letters using ChatGPT. I am making edits after I generate my cover letter, but I am worried that my applications are not getting picked up because of AI-generated cover letters.

What kind of impressions hiring committee get after looking at such a letter? I am in Math.

1
  • 54
    Why are you doing this? Commented Nov 13 at 18:07

9 Answers 9

105

What kind of impressions hiring committee get after looking at such a letter? I am in Math.

For me, the effect would be similar to you writing

I AM A PERSON WHO LIKES TO TAKE SHORTCUTS

in big bold letters in your cover letter. Given that this is the sort of person you are, I would know that if my department hired you, we should fully expect you to continue to use AI assistance to write your course syllabi, committee reports, grant proposals, emails, and many other types of work products.

My likely reaction would be to be grateful to you for disclosing so clearly what kind of a person you are, and for saving me the work of reading the rest of your application and thinking how I should rate it.

21
  • 27
    Whoa, bit harsh. Accurate though. Worth an upvote, anyway.
    – Buffy
    Commented Nov 13 at 20:08
  • 22
    @It'sMe, why hire a human at all, if the AI can do everything just as well?
    – Buffy
    Commented Nov 13 at 21:12
  • 16
    @It'sMe Those are a lot of big ifs that nobody has shown to be true. Commented Nov 13 at 21:20
  • 11
    @It'sMe I would classify AI use into responsible and irresponsible use. Responsible AI use is where you put a lot of thought into writing a document but use the AI to improve your writing quality and/or overcome language deficiencies. That’s fine and is not what I would call taking a shortcut. Irresponsible use is where you try to dishonestly create a superficial impression (which invariably collapses under minimal scrutiny) that you put a lot of thought into writing something, when you actually didn’t. That’s what I call taking a shortcut. And yes, it’s bad.
    – Dan Romik
    Commented Nov 13 at 23:27
  • 14
    @It'sMe also: if taking shortcuts is not bad, then writing on your cover letter that you are a person who likes to take shortcuts should also be not bad. Right? Well, feel free to write that in your cover letters and see where that attitude gets you…
    – Dan Romik
    Commented Nov 13 at 23:30
52

As someone who recently read 20+ variations on pretty much the exact same cover letter for our department's open job, and was able to recapitulate almost that exact same cover letter by inputting a request to ChatGPT to make a cover letter for our job listing, I recommend against this.

ChatGPT is very good for "punching up" or proofreading a draft that you have already written, but left to create letters on its own usually results in pretty generic letters that will sound exactly like everyone else that requested ChatGPT write the letters for them. You want to stand out (in a good way) with your cover letter, and not blend in with everyone else.

1
  • Comments have been moved to chat; please do not continue the discussion here. Before posting a comment below this one, please review the purposes of comments. Comments that do not request clarification or suggest improvements usually belong as an answer, on Academia Meta, or in Academia Chat. Comments continuing discussion may be removed.
    – cag51
    Commented Nov 14 at 22:29
31

This seems like an unwise idea. Meaningless pablum, whether ChatGPT-generated or not, is distasteful to most mathematicians I know, and is unlikely to go unread.

In my experience at a US research university, pro forma letters that are brief and don't say much anything of substance are commonplace and fine; however, I've heard that at more teaching-oriented universities a tailored cover letter is very important.

If you do write a longer, tailored cover letter, then the purpose is to demonstrate that you've put some thought into the position and why it's a good fit for you. My guess is that ChatGPT-generated output is unlikely to convince anybody of this.

3
  • 7
    "the purpose is to demonstrate that you've put some thought into the position and why it's a good fit for you", I'd add that it is indeed a good idea to put some thought into why you think the position is a good fit. This can only be beneficial to the application process. And if the result is a well written cover letter: fine! Commented Nov 13 at 19:04
  • Hello, thank you for the answer. I am editing parts of it, but some words are generic and if I change with other synonym, it just gets weird sounding.
    – Topology
    Commented Nov 13 at 19:37
  • 17
    @Topology How would chat GPT produce a cover letter that explains why you in particular would be a good fit for that position in particular? Editing individual words to synonyms is not the point here, the point is that a LLM generated letter will be pointlessly generic. You need to think about what information you are conveying to the hiring committee about why YOU would be a good fit for their particular position.
    – deee
    Commented Nov 14 at 10:49
15

If you use AI as a tool and carefully edit what it produces, then you should be fine. But, it depends on what your prompt to the AI was and what data it was trained on. If it produces generic text then it won't be very useful and might trigger readers that you aren't a serious contender. It might be obvious to some that it wasn't written by a human, even without a check.

I'd recommend against its use for producing cover letters (and other things), but only as a source of ideas. If I received a letter that I thought was AI generated, I'd certainly have doubts about the sender.

One normally writes a "generic" letter on their own and then tailors it for every application, based on the stated needs of an institution. You don't really need an AI for that, assuming you are applying to a reasonable number of positions. But the tailoring has to be sensible. If it is overly fawning or obsequious then it might be less than helpful, but more likely to be auto-generated.

And, if you aren't a native speaker, then an AI is especially dangerous to use as it may say things that aren't really proper, but that you don't "catch".

3
  • Hello @Buffy: Thank you so much for the detailed answer. I sometimes add specific details to my cover letter, generated by AI (where you ask AI to specifically pull out information from the job Ad etc). How does that sound? The only thing is: the language gets repetitive sometimes.
    – Topology
    Commented Nov 13 at 19:39
  • 4
    That "repetition" is exactly what you don't want.
    – Buffy
    Commented Nov 13 at 20:09
  • 3
    I agree with this. There is nothing wrong in using AI or anything else, as long as the letter is of high quality.
    – Dilworth
    Commented Nov 15 at 1:46
7

It's worth remembering that identifying (or even suspecting) artificially-generated isn't that difficult once you know what you're looking for. To use it successfully, at the moment, for anything much more detailed—or nuanced—than a list or basic description, you'd need to adjust the language so much that you might as well have written it in the first place.

In other words, I suspect that using ChatGPT would either harm your approach or just not save you the time you might think it does.

3
  • 1
    (+1) For me at least (fortunately, a native English speaker), this is what I would expect happen if I did this. However, after having written an n'th draft, I suppose it wouldn't hurt to see what ChatGPT would write and see if it includes anything that I might have overlooked, in which case I'd proceed to write an (n+1)'th draft. Commented Nov 15 at 17:41
  • 3
    And once the AI-generated cover letter is detected, there's a decent chance it goes straight in the trash. Submitting an AI-generated application is basically saying, "Please use your limited time as a human to read my resume, even though I did not find it worthwhile to spend my limited time to write it." No, thank you. Commented Nov 15 at 18:18
  • identifying (or even suspecting) artificially-generated isn't that difficult once you know what you're looking for. - Can you provide a reference? I thought with good prompting it can be hard to tell the difference.
    – Kimball
    Commented Nov 16 at 3:24
6

It is likely rather disadvantageous for you to use LLMs to create your cover letter for you.

The hiring committee will use your cover letter (alongside the other information you submitted) to determine if you are a good candidate for the position. Using a LLM both obscures most of the information (as the LLM can and will - by nature - only generate generic text) while - should they notice that it was created by a LLM - make some likely negative implications about you instead.

If I were part of a hiring committee and notice an applicant submitted LLM generated content, I would need to assume that they don't take the job seriously, are likely to continue using LLMs to generate text for them, or (most likely) both - which would make them a very bad fit.

In short: It is a bad idea, you will leave a bad impression, don't do it.

6

Before I go on I note that the instant I started an answer, the message "Answers generated by artificial intelligence tools are not allowed on Academia Stack Exchange". Bit of hint there, perhaps.

I had a job coach, years ago, who told that the cover letter is the most important part of the application. It's the thing that stops your application being instantly binned (or shredded in more GDPR conscious places).

Every cover letter should be crafted and targeted at the specific role you want.

They won't even glance at your CV if they don't like the cover letter.

One day you'll be the poor sod going through a hundred or more of these and, trust me, AI generated crap will be dumped.

Writing a cover letter (which extends into things like looking for grant money for research and a hundred other things) is a vital skill. Learn it.

1
  • +1 for the last paragraph, which makes a very good point no one else here has thought of making.
    – Dan Romik
    Commented Nov 16 at 21:00
1

Chat-GTP can help with improving an existing draft. It says things things like 'you write this, can you give an example?' or it may suggest better wording (in particular for non-native speakers, it is really good at improving text in this way).

View it in the same way as giving the letter to a friend who may give some suggested improvements, but do not use it to generate a letter from scratch.

0

I'm maybe a bit more pro-LLM than some other perspectives. My feeling is that using an LLM to help you generate text is fine, but using it as a replacement for thought is not.

In other words, a good cover letter should explain why you are a good fit for the position. There is an implicit question here, that requires thought. Which of your qualifications match the description in the letter? What accomplishments do you have that will make you stand out from the crowd? What are your goals for your research and teaching? If all you do is prompt ChatGPT with "write me a cover letter," it will not know the answers to those questions, and it will show that you have put no thought into it.

However, if you have put in the legwork to have answers for those questions, then you can use a language model much more effectively than simply saying "write me a cover letter." You can write a much more specific prompt like, "Write a letter that talks about my accomplishments X, Y, and Z, and that fits into University A which is looking for B, C, D. Make sure to emphasize my experience in super-duper-modern-technology and explain how it relates to Prof E's research in FGH." You could feed it a cover letter that you've written highlighting your skills and the job description and ask it to modify your cover letter to emphasize things in the job description. If you use it well, the result will be something that reflects the thought you have already put into answering the key questions for yourself.

Of course, once you get text from an LLM, you should not simply copy/paste it and call it done. You should read what it has written carefully, and use it for inspiration. Rewrite sentences to put them in your words (you may emphasize some ideas in a subtly different way), correct errors, add pieces that it has missed or that you have realized are important. In the end, you are still responsible for what you submit, and the text should reflect your thought and your work. But you can certainly use the LLM as an assistant to help you express your ideas.

On some level, the thought is the hard part, and coming up with text perhaps "should" be easy once you have done all that legwork. But you might have writer's block, or English may not be your first language, or just want some help since editing is often easier than starting with a blank page.

6
  • 2
    What I don't understand about this approach (as well as about many other uses cases of LLMs that people are talking about) is what's the benefit of it? The time consuming thing about writing anykind of non-trivial text is not typing a few paragraphs, it's (a) coming up with the right contents and (b) expressing those contents in a way that is most appropriate for the purpose and the audience of the text (due to (b) I disagree that coming up with the text were easy). How can an LLM save me a significant amount of time with those things? Commented Nov 15 at 20:31
  • 1
    @JochenGlueck For a cover letter I probably agree this is not going to help much. But, sometimes it is more time efficient or at least more pleasant for me to edit text than to start from a blank page. And I've also been able to use LLMs to take a "vomit draft" that contains the ideas but is too long and informal and make the text more concise, or to convert an outline into a first draft. I think if used properly it's just a tool, and like any tool there's a startup cost for using it that might not be worth it for everyone, and it's not the only way.
    – Andrew
    Commented Nov 15 at 20:43
  • I agree that it's merely a tool. It's just that, with few exceptions, I haven't found the tool particularly useful so far, so I'm trying to understand what other people find useful about it. It might depend on one's personal working style, though. For instance, I would almost always prefer starting from a blank page over editing something that someone (or something) else has written. Commented Nov 15 at 21:01
  • (For the record: I'm not the one who downvoted.) Commented Nov 16 at 12:38
  • I'm not surprised this answer is controversial/downvoted, and lord knows I don't think I'm the source of truth and I could be wrong. But, I would say that LLMs are a big, new technology. One approach is to try to effectively ban a particular use case on moral grounds. Another is to try to accept that people will use the tools available to them and reward effective uses of it. We will see which one turns out to be more successful in the long term.
    – Andrew
    Commented Nov 16 at 13:36

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .