Navigating the intersection of AI and MBA applications is becoming increasingly complex, and this week’s Business Casual podcast, hosted by John Byrne, Maria Wich-Vila, and Caroline Diarte Edwards, addresses a critical component: recommendation letters. With AI tools like ChatGPT providing a tempting shortcut for recommenders under time constraints, the hosts delve into why it’s crucial to ensure these letters remain authentic and handcrafted.

Caroline emphasizes the importance of proactive communication with recommenders. Applicants should engage in open discussions about the specifics of their MBA goals and provide key talking points to their recommenders. By offering a structured brief, candidates can help ease the recommender’s task while ensuring the letter captures unique personal achievements and insights. This approach not only prevents reliance on AI-generated content but also ensures the admissions committee receives a genuine account of the candidate’s capabilities.

Furthermore, Caroline reassures candidates that the writing quality of recommendation letters isn’t under scrutiny. Business schools focus on the substance of the feedback rather than grammatical finesse, which is especially relevant for recommenders who may not be native English speakers. The key is the conveyed messages, not the linguistic style. By managing these elements, applicants can secure impactful and authentic recommendations that strengthen their application without falling into the AI trap.

Episode Transcript

Note: This transcript was generated by AI and may contain minor inaccuracies.

[00:00:06] – John

Well, hello, everyone. This is John Byrne with Poets & Quants. Welcome to business Casual, our weekly podcast with my co-host, Maria Wichvilla and Caroline Diorti-Edwards. We’ve talked in the past about the dangers of using AI to write your essays for your MBA application. But one thing we haven’t addressed that seems to be occurring these days, is a recommender who might go to ChatGPT, insert a few facts about you, and then have ChatGPT do the recommendation letter. This is happening, and it is problematic. We want to basically address the issue and give you some advice as to how to handle your recommenders and prevent them from using AI in a way that can be detected by the admissions officials. Now, interestingly enough, Caroline had written about this quite some time ago. Of course, she is a co founder of Fortuna, and her colleagues write for us on a regular basis. Back in 2023, when ChatGPT was really new. Not much has changed since then, except that ChatGPT has gotten better, and therefore, the temptation to use it by a recommender who is obviously busy at work may be even increasing. Priest. So what do you do?

[00:01:32] – John

How do you tell your recommender, please do not use artificial intelligence to write this letter because admissions officials will sniff it out and it will cause me harm. Caroline, what’s your general advice as to how an applicant should handle this?

[00:01:50] – Caroline

Yeah, well, definitely. I advise candidates to address this upfront, and they should sit down with a recommender at the beginning of the process and discuss with them their MBA application, why they’re applying, their goals, brainstorm together on the messages that the recommender could include in their recommendation. And they should have a proactive discussion then about how it’s important that the recommendation is in their own words, and it’s best to avoid outsourcing the writing to AI, because admissions far readers are very adept at picking up on inauthentic writing. So first of all, be proactive in addressing it rather than reactive if you see that that’s what they’ve done. And then if you provide them with a good brief, so we advise candidates on having that discussion, but also putting it down on paper and not writing everything out for them, because then they might just copy paste what you’ve written. But give them some bullet points with some of the key talking points, because it can be very hard for a recommender to remember everything that you’ve done that’s made you stand out. They’re probably managing many people. They don’t necessarily remember all the details and the ins and outs of what you’ve done and the particular role that you had versus someone else and the impact that you had and all of those details.

[00:03:14] – Caroline

I’ve written recommendations in the past where people have not given me a brief. I remember scrambling to go back through old performance reviews where I had documented what that person had done because I couldn’t remember everything. It’s a pain to have to go back and do all that. So it’s good if you can do that donkey work for them and give them some good talking points, but keep it in summary format so that they have to put it in their own words. So if you can give them a really good brief, then it does save them a lot of time, and it should be then pretty easy for them to put it together in their own words. I also advise candidates to talk with their recommenders about the fact that schools aren’t concerned if the recommender can write really well. So of course, schools like to see polished writing from students because being able to write well and get your ideas down on paper is a good skill to bring as an MBA student and may be useful for you in your future career, but they are not evaluating the writing skills of the recommenders, right?

[00:04:27] – Caroline

And I would often see that at INSEAD where we had a very internationally diverse pool, and students and recommenders would sometimes be concerned that the recommender was not a native English speaker, and their grammar might not be very good, their spelling might not be very good, they might have some awkward terms of phrase, and they would be concerned that that could be damaging for the candidate. And that’s absolutely not the case at all. The school doesn’t care less about the grammar of the recommender What they care about is the context, right? They’re looking to get the key messages and understand the feedback of the recommender, and they really don’t care about the… Whether the writing is stylish and beautifully written. It’s really just about getting the messages across. I think you can also reassure the recommender as well that it’s not a writing competition for anybody, and it’s especially not a writing competition for the recommender.

[00:05:31] – John

Maria, let’s go back to maybe a question I didn’t ask Caroline, which is actually even more basic than what should you do as an applicant, which is why shouldn’t the recommender use ChatGPT? After all, if I take a few snippets from the brief that the applicant provides me and I insert it into the prompt, ChatGPT is shockingly good at composing a very smart thoughtful essay, or in this case, letter, with those details embedded in a way that actually could be quite compelling.

[00:06:10] – Maria

Yeah. First of all, I think one of the key things here is that you have to, at a minimum, have some of those specific examples to feed into the AI. Because if you just say, Okay, I have this young employee who’s applying to business school. Give me a letter of recommendation, that’s where things are going to be really suboptimal because it’s just It’s going to be a very generic letter. It’s just going to say things like, Oh, they’re very nice, or they’re hygienic. They show up on time. It’s not going to have any of that specificity. The whole point of the letter recommendation in the first place is to get those specific anecdotes and to get that very particular perspective of what do your superiors think of you. And so if the letter does not provide that because it was just spat out by an AI, then it’s not at all going to meet why these things even exist in the first place. And then I think the concern with even saying, Well, okay, Maria, but I’m going to give them the specifics, and then ChatGPT is going to spit it out. Chatgpt has a very specific voice to it, and maybe you don’t notice it because you are an applicant in your little bubble, and maybe you see some ChatGPT from time to time, maybe you don’t.

[00:07:21] – Maria

I can assure you that admissions officers, especially by this point in the game, I don’t know that their accuracy is 100 %, but I can assure you that they can They can probably spot an AI-generated written anything, whether it’s an essay or recommendation. They can start to spot it from a mile away. There are certain, I don’t know, linguistic ticks, for lack of a better term. When something is AI-generated, maybe you can’t always tell, but there’s an uncanny valley, like an uncanny aspect to it that used to plague a computer animation generations ago, right? This idea of it doesn’t feel human. And so if it doesn’t feel human, it’s not going to feel authentic. And again, if I were an admissions officer, I wouldn’t bother asking for a recommendation if I didn’t think it were authentic. It completely defeats the purpose of asking for them. So to the extent that if everyone starts using ChatGPT, okay, I put in my bullets and then it spits something out, does it not stand to reason that every ChatGPT generated recommendation is going to sound very similar? And that even if those patterns haven’t been obvious up until now, they’re certainly going to start being obvious pretty soon.

[00:08:30] – Maria

If every single recommendation follows you, a three-paragraph structure, and every single recommendation ends with, I heartily recommend this candidate, you’re going to start noticing. So as with anything, I just feel like I get the temptation to take a shortcut. No matter what that shortcut is, the shortcut is always the tempting route in any facet of life, but the shortcut is usually the wrong or at least the suboptimal route to take.

[00:08:57] – John

No, Maria, when you mention Hygienic, I’m thinking ChatGPT saying, This fellow comes to work showered every morning.

[00:09:05] – Maria

He smells terrific. I always use that as a sarcastic example of sometimes when people are like, Oh, my recommender is going to talk about how good I am at solving differential equations in my head. I’m like, You’re missing the point. Nobody cares if you’re… They’re very punctual. Their email signature file is really nice to look at. Like, What?

[00:09:29] – John

You When you’re right about ChatGPT literary tics, I mean, the most common literary tic, frankly, are bullet points and dashes. Whenever you ask ChatGPT anything, it bullet points you to death. The same thing with the dashes, they’re all over the place. It’s uncanny the way it forms a reply of one kind or another. You’re right, even if you do submit some specifics to allow GBT to hallucinate over them, the problem is there’s a commonality in the way that ChatGBT would respond. The other issue here, in addition to just an admissions official sniffing out a problem, there are admissions officers that are using software to detect ChatGPT and other Gen AI platforms that have been used for essays, primarily, but also, I would imagine, for recommendation letters. And yes, not all admissions officers use this software, and there are a lot of false positives that occur with it. But the software is getting better at detecting this stuff. So it doesn’t even have to be, Oh, it doesn’t just smell right or it doesn’t look right. There could be some software involved in here that could get you in trouble. But then the issue is, isn’t it offensive for an applicant to tell a recommender, Hey, don’t use AI?

[00:11:05] – John

Or is there a subtle way? Like more admissions officers are using software to detect AI written content. So please try to, I don’t know, exercise your own thoughts about me and write it your heart or something like that. What would you advise, Caroline?

[00:11:36] – Caroline

It’s definitely something to have an open discussion with them upfront. I think the way that you put it in, giving some background explaining the risks if they do use AI, hopefully they will understand that. It may be worth, even before you commit to having that person as a recommender, it may be worth sounding them out about their willingness to write for you and their willingness to write it themselves rather than use AI. Perhaps you want to have a more informal discussion out them first, a more informal discussion with them first to sound them out and get a sense for whether you think that they will be truly invested in the process or whether you think that there’s a risk that they’re going to outsource it to AI. And get a sense for whether they will really put their heart into it, as you say. And if not, then perhaps that’s not the best choice for you, because the worst case scenario is that you don’t have that discussion with them upfront, and they don’t do a very good job with the recommendation, and maybe you’d never find out about it. You might get dinged and you’ll never know what happened with your recommendations, or they might send it to you for a review and you see that it is written in a very generic fashion, and it’s more difficult to have that discussion at a later date or even switch recommendations at a later date.

[00:13:11] – Caroline

That’s awkward. So definitely better to sound them out up front and really explain to them your reasoning and the risks of using AI.

[00:13:23] – John

I’m also thinking that the easier you make it for them to write an original recommendation letter, the more likely is they’re not even going to be tempted to use AI, and that comes back to the brief. Maria, in terms of a brief, how detailed should it be? How long should it be? How does it come off not as coaching in a way that’s a little too forceful or aggressive from the candidate’s standpoint.

[00:13:47] – Maria

Yeah. So this is actually one of the indirect benefits, in my opinion, of you having the recommendations as well, is that it is an indicator to the admissions committee, albeit an indirect one, of your ability to manage upwards. So So if somebody submits a recommendation and it’s lukewarm or even negative, wow, what does that say about that person’s self-awareness? What does it say about their ability to influence those over whom they have no direct authority? So right off the bat, so How do you manage it? Well, it’s going to depend on your relationship with that recommender and you knowing how open are they to feedback? How open are they to coaching or to following your lead on certain things? So I think that the specifics will vary based on your relationship with that person. But I think you can frame it as, first of all, you’re really busy. Thank you so much for doing this. When you’re asked certain things, here are some things I thought you could talk about. So when you’re asked, talk about a time that you gave the candidate critical feedback and how the candidate responded, obviously, you can write whatever you want.

[00:14:49] – Maria

But remember that time that you told me that I needed to take more showers and I wasn’t hygienic enough? Now is your chance for you to whatever it might be If you phrase it more as like, You can write whatever you want, but I thought maybe you could talk about the time that I did X or Y. I think it is important. I don’t feel that it’s inappropriate to put this level of coaching, especially I think sometimes our bosses tend to focus on how good we are at doing our current job in the moment, which by definition for most MBA applicants is a lower to mid-level type of job. If somebody is a computer coder, they’re The recommender might be tempted to say, Well, this person’s code is constantly bug free, and their semicolons are always exactly where they need to be in the code. But that’s not the quality that an admissions officer is looking for. I actually have no problems at all. I have no ethical qualms whatsoever with that coder going to the recommender and saying, Hey, I know that you think that I’m great because my code semicolons are always flawless, but it turns out that these admissions officers don’t really care about that, and here’s what they do care about, so please We talk about X, Y, or Z.

[00:16:02] – John

You mentioned a word bug, and that makes me think that one of the bugs involved in this process is a recommender who doesn’t follow through on deadline for an application. Since these deadlines are coming fast and furious right now, how do you gently nudge a recommender who hasn’t yet filed the letter with the school. Caroline, your nudging technique is what?

[00:16:35] – Caroline

Well, I think you need to manage the timeline with them well ahead of the deadlines. Ideally, they’ve submitted a week or two ahead of the deadlines. But if you’ve got to that point where they haven’t, then I think you just need to explain to them the stakes for you and ask them when they think that they can submit it by and get them to commit to a time frame and follow up with them and just maybe resend the brief to them. Maybe you sent it to them three months ago and they’ve lost it in their email. Perhaps there’s something that you can do to just make sure that they’ve got everything that they need. They’ve got the right links from the school. They’ve got the brief. Maybe you can ask if there’s something you can take off their plate so that they’ve got time to work on your recommendation. It’s tricky, It’s stressful because you are not their boss by definition, yet you’re trying to manage upwards and get them to do something when you have no control over the time. So it can be very stressful for candidates. But that’s why it’s best to, even in some cases, give them a false deadline of a week ahead.

[00:17:50] – John

The false deadline trick.

[00:17:52] – Caroline

Yeah. We do with my father-in-law, we always tell him the airplane is leaving an hour before it’s actually leaving, so he doesn’t miss the airplane. So just to make sure that there’s some buffer time built into your timeline.

[00:18:07] – John

Maria, do you have any other nudging tips?

[00:18:10] – Maria

I mean, I think, presumably, the person who is writing your recommendation is someone who is in your corner, who is heavily invested in your success, who is rooting for you very, very strongly. If they’re not, then maybe you should, as Caroline mentioned, maybe you should re examine your choice of recommender. But so presumably, if that person is rooting for you very strongly, I think it’s perfectly fine to say to them, Look, if you don’t get this in by the deadline, my candidacy might not be considered complete. I might be pushed to the next deadline, right? Or if we’re talking round Round 2 for a school that doesn’t have a round three or round three for a school that doesn’t have a round four, I might not be able to attend this year at all. So I think, as Caroline said, if you explain the stakes to them, I do think that that can be the thing to help to help nudge them over that finish line.

[00:19:03] – John

Okay, so there you have it. Make sure your recommender doesn’t use AI to write your recommendation letter. Make sure you give them a deadline that works for Caroline’s father-in-law, that it’s super early so that even if they do get into a little delay or postpone, it’ll still get there on time. Nudging is a good thing. Don’t be afraid to nudge and don’t be afraid to write a very detailed briefing to make this job easier for your recommender. This is John Byrne with Poets & Quants. Hey, good luck on those applications that you’re submitting over the next few weeks. We’re rooting for you.