How might AI tools such as ChatGPT make the MBA application process easier? Can you use ChatGPT for your business school research? Can you use it to help you brainstorm for your essays? How about for actually writing your application essays? And should you use AI for your application at all? In this episode of the mbaMission Podcast, Harold Simansky, Jessica Shklar, and Jeremy Shinewald discuss when you should—and should not—get assistance from AI with your business school applications, and how to do so effectively.
Each week, Senior Consultant Harold Simansky takes a deep dive with some of his mbaMission colleagues into various aspects of the business school admissions process, from how to select the right schools to choosing and managing your recommenders to drafting compelling application essays and preparing for MBA interviews. A brand new episode of The mbaMission Podcast is released every Tuesday, and you can find it on all the major podcast streaming platforms, including Spotify and Apple Podcasts. More of a visual learner? Check out the mbaMission YouTube channel for full episodes.
Harold Simansky: Hello, Jeremy. Hello, Jessica. Let’s talk about artificial intelligence. First, a disclaimer: this podcast is not created using artificial intelligence. This is natural intelligence right here. [Laughs] Okay, Jessica, I’m going to throw it right to you.
Jessica Shklar: I think when you think about it from an admissions perspective, from the applicant perspective, AI is a tool—or maybe a better way to say it is it’s a really good starting point, or it can be a good starting point, but it should never be the ending point. And by that I mean you can use AI to help you come up with ideas, to help you get started with a draft, to help with some editing, to find interview questions, to do basic research.
Harold Simansky: Yeah, do basic research.
Jessica Shklar: But if you stop with what AI does for you, you’ll never, well, I think you have a phrase that says, “AI can help you be mediocre.”
Harold Simansky: That’s right.
Jessica Shklar: So it’ll help you be mediocre. And you just don’t want to be mediocre in this process. I’ll start with where we’re going to end, which is [that] the bottom line is that AI is a terrific tool, but it is not a substitute for your own thinking.
Jeremy Shinewald: We can break it down even further and say that, let’s say we use cases where the documents or whatever your output is is ultimately admissions committee facing and those where it is not. Harold, which one should we go with first, admissions committee facing or not?
Harold Simansky: Actually I will have many clients start using ChatGPT for research, and to sort of pick up on something Jessica said beforehand, if I go now to buy a new TV, ChatGPT is going to be my go-to place. How does that affect business school? Are there similar decisions made along those lines?
Jessica Shklar: Right. And I think we have really in-depth Insider’s Guides that go far deeper into each school, but it can take a lot of time to read them if that’s your initial phase. So to say to ChatGPT, “Give me the top three differentiators between Harvard, Stanford, and Wharton” or, “What are the most popular classes at these schools?”—that’s a good comparison tool. And then I would use our Insider’s Guides maybe to go much more in depth.
Jeremy Shinewald: I would say the one thing that’s important is, you know, the mediocre outcomes of it. You can say, “ChatGPT, tell me about the top six strategy classes at Wharton.” Okay? You have to go back and check to make sure they actually are classes. Like, if you’re using that, maybe you’re putting it in your paper, and maybe that professor left and [the class] hasn’t been offered for six years. Maybe ChatGPT made it up.
Harold Simansky: Absolutely.
Jeremy Shinewald: You know, you always have to be skeptical of something that’s out of your control. But I agree, the research piece can be quite powerful.
Jessica Shklar: Tends to be more surface, more comparison, more of an overview, and then no substitute for our Insider’s Guides or for going to visit a school or for talking to students.
Jeremy Shinewald: We put together onTrack, we have five, six, seven, eight different videos on each school, and none of them were designed by ChatGPT. They were all done by the research of a human being. They’re all accurate. There’s no question that if you reference something, then you still should personalize it toward what you’re doing. You wouldn’t say, “I outsourced this research to a friend.” Or, “I outsourced it to a research assistant.” That’s what you’re doing with ChatGPT. There is a role [for ChatGPT], but you’ve got to be careful.
Harold Simansky: No, and listen, the reality is, it also lends itself to opportunity. I’ll tell you exactly what I mean. If everyone is using ChatGPT to do research, many people should be, but they do not use our Insider’s Guides—which are free; they’re on our website. We have a lot of things going on on our website. Please check it out. All of that noted, if everyone is doing research using ChatGPT, what a premium there must be to talk to somebody, to actually have a live person, have a quote in there [in your essay]. Can you imagine the admissions committee? They’re sitting there, reading essay after essay, they sound exactly right, because it’s all written by ChatGPT, and then your essay comes up, and you’ve actually spoken to somebody. Maybe in fact you have a fun essay, and ChatGPT is many things, but it is not fun.
Jessica Shklar: So we start with research, and then let’s talk about our own internal research, which is brainstorming. As we think about essays, instead of diving right into the questions, is there a role for ChatGPT in brainstorming? And I would say yes, but. It takes a lot of input for ChatGPT to get to know you. They’re not sitting at your dinner table. They’re not hovering over your shoulder. They don’t know who you are. So if you said to ChatGPT, “Give me some ideas for what I could write for ‘what matters most to me and why’ for the Stanford essay,” well, they will probably come up with five or six generic topics. That doesn’t mean that’s what matters most to you, or to you, or to me. So then you have to take your own idea and say, “This one actually does resonate. How? And what are examples from my own life?” So again, ChatGPT can get you started when you’re frozen with ideas. Before writing, which I think is going to be where many of you listening to this think about using ChatGPT, let’s think about brainstorming.
When you sit down with five essays or three essays for a school, and you don’t even know how to think about them, maybe that’s a place to start with ChatGPT. Say, “What are some ideas?” Or even, “Here’s my resume; what are some ideas?” But again, that’s very limiting, because ChatGPT is not a person who knows you, who talks to you day in and day out. They’re not your friend. They only know the little bit that you give them. It can trigger you to start thinking. I think it can get you unstuck, give you ideas of what other people have done, but then you have to overlay your own abilities and knowledge of yourself.
Harold Simansky: Right, right.
Jeremy Shinewald: I think when it comes after you’ve got your outputs, and you really know what you want to say, I think it’s ironic. You need to take a massive step back by asking ChatGPT to write that, write me an essay about how hard it was to start an MBA admissions consulting business, you know, the first few months of that. It’s like, that’s, you can get from something where you, the juices were flowing to something totally, totally generic. One of the ironies I think of as I’ve seen people submit things through ChatGPT that are clearly ChatGPT is that it could be anyone. It’s so vague. It’s got a certain vagueness. And then every once in a while, someone will say to me, “Well, I started with ChatGPT.” And I’m like, “Well, this is so specific now. You’ve done so much revising and changing that ChatGPT basically wasn’t there in the first place.” It was like it served almost no purpose for you—maybe giving you a rough structure of how to get started.
Jessica Shklar: I think what it does do with the writing is help you overcome writer’s block. Sometimes, as you know, how often do we just throw away the entire first paragraph in a client’s essay? Now maybe they’re taking that first step. They’re getting something generic from ChatGPT, but it’s so much easier to edit than to create. And so maybe they’re using that as their first draft to get them unstuck.
Harold Simansky: That’s right. I actually want to pause here, because I think it’s an important point, and that is good essays, good applications, start with introspection and brainstorming. As a firm, we spend a lot of time on that. Can ChatGPT help you with that? No, it simply can’t.
Jeremy Shinewald: Yeah. It’s not there to have creative thought. It’s there to reflect other people’s thoughts, and that is not your own. But what about other positive use cases? Let’s be fair to technology. Other use cases, have you had applicants who, I don’t know, maybe said “You’ve got to truncate this five-line bullet point down to two lines for a proper MBA resume.” You could put this through ChatGPT, get it down, break it into two bullet points, dabble.
Jessica Shklar: Then you have to look at it and make sure it’s accurate and impact focused. I did this recently with one of our internal documents. You guys don’t know me, but I’m long-winded and detailed, and Jeremy’s not. And so we sometimes butt heads on this. And he said to me, “This is four pages, get it down.” And I couldn’t, because I like to throw every single detail possible into everything that I do. And so I put it in ChatGPT, and I said, retain all the key bullet points, but make this shorter. And you know, it got about 85% of the way there. It did a really nice job.
But it didn’t know all the nuances. And I had to then do a line-by-line comparison with our original document to make sure they hadn’t missed anything. So it was great. It got me started on something that I was struggling with. But I still had to spend several hours fixing it.
Jeremy Shinewald: That’s a great example, I think, of the generic nature versus the specific nature, but the generic still having some use. So again, giving credit to technology here. We have a massive, massive bank of interview questions from every single school and every single part-time and EMBA program within those schools, and it goes on and on. And we’ve just been harvesting these questions for years and years and years. And again, even those, we have to go through and make them, you know, make them specific to our candidates. We can’t just say to anyone, “So what was it like to have, you know, six years of manufacturing experience and suddenly transition to strategy consulting?” You know, even we have to do that. If you wanted to start some basic interview prep, you could say to ChatGPT, “Give me a list of a hundred questions that an MBA might be asked in an interview.” And I don’t think it would be necessarily indicative of the type of experience that you would have, and they wouldn’t be that specific, but it’d be enough to start practicing, enough to get going.
Jessica Shklar: And that just keeps coming back to the same point. ChatGPT is a really, or any AI, is a really good starting point, but not an end point.
Jeremy Shinewald: If you were to only use AI-generated questions, you’d end up at mediocre.
Harold Simansky: That’s right. Listen, ChatGPT does some things very well. It takes a 2,000-word document, and it can turn it into a 1,000-word document. Accurately, yeah, really pretty good. That doesn’t mean that you’re done with that, but really, it’s a great place to be. I’d rather edit 1,000 words than 2,000 words. That’s certainly the case.
Jessica Shklar: Right, but you still have to make sure they didn’t cut the wrong thousand words. I remember when we were first hearing about AI in applications. Jeremy, you at a staff meeting presented an essay that you had asked ChatGPT to write, and you had given it pretty specific guidelines and sent it to us. It was pretty good, but it wouldn’t have gotten anyone into business school, because it was not fun. It wasn’t personalized.
Jeremy Shinewald: I was trying to sit there with ChatGPT for a while and get it to the level of detail that I like to see in a “why Stanford,” a “why Wharton,” a “why Chicago,” and really push it. Maybe I’m just a dinosaur, and I don’t know how to push it appropriately, but you know, “Can you add more resources? Can you add another class? Can you add a club? Can you add this?” And it just doesn’t read in a compelling way. It doesn’t read like a human being who really knows how they’re trying to put their argument together for why they really fit [with the school]. Again, there’s that word—fit. A snapshot of a million Googles doesn’t fit.
Jessica Shklar: You have to fit. And that’s perhaps the easiest kind of question to answer for ChatGPT, because it’s research based. Stanford’s personal background [question]—How has your personal background impacted a recent decision or action you made in the last few years?—well, ChatGPT won’t be able to do that, because they didn’t grow up with you. So it’s struggling to be beyond mediocre for the easiest question.
Jeremy Shinewald: One other place, let’s say a school had a super hard character count or word count on an essay, which happens from time to time, and you’ve got to cut out a few words. Yeah, you can use it there. Let’s say you wanted to do a copy edit, but like I would even say in circumstances like that, some of the blemishes help you. In talking to admissions officers, they don’t want to see something that’s smooth. I remember talking to a Harvard Business School admissions officer who said to me that she doesn’t really, she doesn’t want to see sloppy, but she just doesn’t really care about a semicolon instead of a colon. Doesn’t bother her. It’s part of the flavor of an applicant, and especially if you’re an international applicant and English is your second language. You don’t need a computer to take out the idiosyncrasies of the way you write or speak.
Jessica Shklar: I would make it even stronger than “don’t need to.” When I work with international applicants, I promise them that everything they submit is going to be clear and that they’re not going to be contradicting themselves or using words incorrectly. But do I promise to make them sound like a native speaker? That would be against their interests. I certainly wouldn’t want ChatGPT to polish an essay beyond where it would make sense, because remember, let’s say you get to the interview phase with a school with those essays, and then you get to an interview, and it’s clear that your English, as good as it may be, is not at the same level as your essays. That’s going to be a disconnect for the admissions office.
Harold Simansky: That’s right. Listen, in my mind, ChatGPT allows you, almost forces you, to be interesting. If you think about it, if sort of the default now is going to be sort of mediocre, like I’ve heard it before, that syntax is certainly going to be very similar. Ernest Hemingway is not popping out of ChatGPT. Foster Wallace, there is no ChatGPT. Right? So the reality is, how do you become interesting? And that really requires a lot of work, a lot of introspection. And again, things that ChatGPT will not allow you to do. It’ll trick you in some ways that you think you’re doing a great job. It’s a terrific essay. I hit word count. I didn’t have to use the magic hyphen.
Jeremy Shinewald: The other thing also, maybe this is a transition to our second part if we’re ready, but schools are running AI checks. They are thoughtful about this. They’ve been running checks on recommendations for years and years to make sure that applicants aren’t, in fact, in some ways they’ve been ahead of AI in that regard. They’ve been running people’s recommendations through checks to see if the voice is their own or their recommenders’. And so, you know, they’re checking, and if you’re using something that is substantively still AI, you have a serious problem.
Jessica Shklar: That actually raises a question that our clients and any applicants might be interested in is, how should you advise your recommender about using ChatGPT or not?
Harold Simansky: I actually have some thoughts about that, because over the years, what I’ve certainly seen is recommendations becoming more vanilla. The reality is, for all sorts of different reasons—one of which now is AI, but other reasons as well, as the entire process has become more systematized, as more people have been applying, as long as there’s been a history of people writing recommendations—they’ve become super vanilla. So at the end of the day, what I’m really encouraging applicants to do is just like push much more to the recommender.
Basically say, “I’m going to give you some ideas about what I’d love for you to talk about,” but at that point, don’t write anything for them. Just push it to them, push it to them. We’re looking for authenticity. And I think that’s really the case with essays. I think that’s the case with ChatGPT generally. Authenticity is the name of the game right now.
Jessica Shklar: We can be really, really directive for applicants out there. If your recommender comes to you and says, “I’m going to use ChatGPT for this,” your pushback should be, “That’s a good place to start, but unfortunately, business schools are alert to that, and we’re starting to see that those recommendations become quite generic. So if that’s where you want to start, that’s fine, but could we then meet and go through specific examples or anecdotes that you could put in it to make sure that this is in your voice and really reflects me as a candidate?”
Jeremy Shinewald: I would even go further and say, “I really want you to have an efficient process with this. … I’d like to review some of my major accomplishments with you. I’d be happy to furnish you with a list of them. We can talk about them, but I would really appreciate it if you’re here to support me, if you could write this whole thing from scratch and just, and that would really reflect you.” Now I understand that some, can’t depend on people, but I think maybe the opposite way of saying that is, “If you are going to get to the point, or if you are not going to get to the point where it’s largely your own original output and where ChatGPT is still evident, it’s going to hurt my candidacy. Please don’t write my letter of recommendation.”
Jessica Shklar: And look, to the applicants listening to this, we know that’s a hard conversation. You want to come on to a free consultation with us and role play that conversation, we’re happy to help you with it. But it’s hard, and we get that. Sometimes you just have to do hard things.
Harold Simansky: Now sort of moving forward in the sense of what’s happening inside the admissions office as far as, how are they using artificial intelligence? Jeremy, you started talking about it as using it simply to sort of assess who wrote what, but are there other ways, maybe even more sophisticated ways that they’re using it right now?
Jessica Shklar: I think I’m actually going to switch that question around and say that what we’re hearing from admissions officers when we’ve spoken to them and from what we’re reading is that they’re doing what they can to avoid using it more by challenging applicants to avoid using it. And we see this with questions that have many more subparts that really force you to be specific, or video questions. MIT, your alma mater, has pop-up video questions that you cannot prepare for, because they’re trying to test your spontaneity, test your ability to think on your feet. So, ChatGPT can’t do that for you. They can’t even prepare you for that. We can.
Jeremy Shinewald: Dovetailing on what Jessica’s saying, let’s think about Harvard Business School for one second. About a decade or so, they had a very long prompt, then they cut it back to 900 words. It was almost, like, effectively, “Tell us anything you’d like us to know.” It was more, very, very open-ended.
Jessica Shklar: Very open-ended.
Jeremy Shinewald: That is the perfect kind of essay to be completely abused and ruined in an AI, in a ChatGPT world, right? Like I imagine that they foresaw that that question was just effectively out of date. It was like a dinosaur, because it was going to invite vague answers. I don’t think they were afraid of ChatGPT. I think they were afraid of not getting interesting data, not getting interesting stories. And so they changed it to three very short, extremely detailed and complicated, some of these vexing prompts that you can read them several times and still kind of be like, what’s going on here? What do they want here?
Jessica Shklar: They don’t naturally fit together all the time as well.
Jeremy Shinewald: And there are transitions in like areas in those questions like how did this, how did this experience impact your perspective on this going forward? Like they’re all extremely cumbersome and confusing, because they don’t want individuals to be able to have this big, kind of open opportunity. They want people to force them into a box. I bet you they might even, maybe they’ll last a year or two, and then they’ll change, because they’ll be saying, “Oh, there are examples online. We’re starting to get the generic versions of these.”
Jessica Shklar: And two of those essays are only 250 words with multiple parts. It really forces you to avoid being generic. Every single word in there has to be specific to you and add meaning.
Harold Simansky: Jessica and Jeremy, this is not a bad time to pause and say we as a firm have spent a lot of time sort of figuring out the HBS essays. There’s three of them now, as you said—300 words, 250 words, 250 words. They’re very hard. At the end of the day, they are very hard in a way that other essays weren’t, as well as certainly HBS in the past wasn’t. And I have to tell you, through all of our work, many, many hours of work on all of our parts, and as our entire firm really looked closely at HBS essays, none of us could use ChatGPT to figure it out. The reality is, it’s very personal, it’s very hard, and there is no role for ChatGPT in an HBS essay.
Jessica Shklar: And then I think MIT [Sloan] is another example of a very different way. They have not really changed their application that much, but it involves a cover letter and then a video and then two pop-up, well, one pop-up video question, maybe two.
Jeremy Shinewald: A recorded video that you can do on your own and then also a pop-up video, like, you’re just gonna get a prompt, and you have to respond on the spot.
Harold Simansky: You know, in fact, let me use this to take a step aside and talk about the one-minute MIT video. If you may remember in the past, there was really no requirements around it. Give us one minute. Over the last few years, they said, give us one minute now, on your iPhone, one take. One take, yeah.
Jessica Shklar: No transitions, no music.
Harold Simansky: Exactly. Because back in the past, you can find them on YouTube, it was like these Academy Award–quality videos.
Jessica Shklar: It was a creativity exercise, and that’s not fair, because business school is not about how well you know how to do video editing.
Harold Simansky: Exactly. And that’s the case in my mind with ChatGPT. It’s anti-authentic. Is that a term? ChatGPT is anti-authentic. We see from business schools, in all sorts of different ways, the use of video interviewing, that now they’re looking for authentic. So this notion here of ChatGPT will make you a better applicant is simply not true, because it’s going in the wrong direction. You’re more inauthentic. You’re more artificial—exactly what business schools don’t want.
Jessica Shklar: And so, I think that admissions officers tend not to be of the generation that grew up using ChatGPT. I’m sure some people are, but for the most part, they’re not. And so, that’s not their orientation. They’re not going to think about, how can we use this? They’re going to think about, how can I get authenticity from applicants so I can really get to know them? Because their goal hasn’t changed. Their goal is to build a diverse class that reflects [multiple facets] and that makes sure everybody at the school has an amazing experience. That’s not going to happen if you can’t read who someone is because they’re too generic.
Jeremy Shinewald: So as a trend line, I think we can be saying, you know, we’re, we would expect to continue to see shorter essay prompts, maybe a little more complicated. We would expect to see more video prompts, which I think has been a trend. Like there used to be one school or two schools, like Kellogg was first, then Yale and Rotman and MIT. Yeah, there are a few others. Or some of them have video interviews as well, which are effectively spontaneous prompts, shorter videos, others, and other spontaneous things. Like, I wonder if they’ll be, you know, “We’re gonna call you at some point in the next like 12 hours, you know, be ready,” or, you know, “Are you free? It’s the admissions committee calling.” … I would expect that we almost, this is significant conjecture here, but nonetheless, I would think that we might almost get to the point where there’s like a rough cull, and then it’s like, “You’re all invited to campus, you know, for a day,” like almost like a Super Sunday day for investment banking or something like that.
Jessica Shklar: The Wharton Team Based Discussion interview was actually like that, if you think about it in sort of a proxy form. And for those who are not applying to Wharton or are not familiar, instead of a one-on-one interview, for Wharton, you actually meet now, it’s in a virtual room. It used to be on campus or some other satellite locations, and four or five of your closest friends, four or five other applicants, you’re solving a business problem together.
I understand that it sounds terrifying. It’s not a complex business problem. You won’t need a calculator. Or a spreadsheet. In half an hour. It’s more a problem—like, develop a course around these parameters or create a day of orientation. So it’s not a business problem. I think it’s a little bit scarier than it actually is. But there’s no way to prepare for that using ChatGPT. And once you’re in that session, you’re dealing with the variables of five other people.
Harold Simansky: That’s right. Jessica, Jeremy, I’m sorry, I have to pause right here. You know who does a great job of preparing people for the Wharton Team Based Discussion [TBD]? It is us. We take six of you who’ve been invited for an interview. We put you in a virtual room, and we actually run it just like you will see if you get a Wharton TBD.
Jessica Shklar: I will say that, although from an admissions perspective, they’re not the most dynamic thing that we ever work on with our clients, it is my favorite thing because of how completely transformative it is. You see these applicants walk in, and they all have this look of terror on their face. An hour and a half later, they are walking out confident. It is the fastest and most dramatic transformation we see in the admissions process. I love it for that reason.
Jeremy Shinewald: I see people who go in, and it’s like everything is about deference, and, you know, like it’s not that we’re telling people to be impolite, but no one is prepared to advocate for ideas, because they’re all afraid they’re gonna seem like they’re being too alpha. And it’s like, you get to the point where it’s a natural discussion, and that’s where you need to be.
Jessica Shklar: And in fact, I remember when they used to do these in person, and ours were not quite the same, because they were virtual. And then when COVID happened, Wharton decided to keep those Team Based Discussions virtual, which means our preparation for it is now exactly the same as the ones there.
Harold Simansky: Right, right, right. Absolutely. And again, to get back to theme here, that is not anything that ChatGPT can help you with. Not at all. It’s simply artificial intelligence. Again, it’s not artificial intelligence, it’s natural intelligence. And that’s really what business schools are looking for.
Jessica Shklar: So I said at the beginning, we were going to end at the same place. And I don’t think we’ve evolved from that. I think that’s exactly what we did. ChatGPT or AI can get you to mediocre. The AI tool is a really good place to start, but it is meaningless if that’s where you finish.
Harold Simansky: Agreed.
