Dos and Don’ts of AI Usage in the MBA Admissions Process

The point of an MBA application is to share your story with admissions committees, and essays are the space where your voice can most come to life. In recent years, AI, while an incredible tool, has introduced a major risk for MBA applicants. AI creates writing that is deceptively polished but not deeply reflective. It can say a lot, without really saying anything at all.

Clients often ask me when it is appropriate in our process to use AI. AI is a great tool for backend tasks, not front end. You can think of AI as your research assistant or administrative helper. For instance, clients and I use AI for call notes, and I often use AI to query those notes. We also sometimes use AI to supplement school research, however, everything you find there should be fact-checked. For example, you could ask ChatGPT to tell you the most popular leadership courses at your target schools, and then you could use that knowledge to read about those courses online, and to ask students and alumni about their experiences in those courses. What you do not want to do, however, is plant those course names into your essay without doing any research in between. The course could no longer exist, for instance.

I often joke with clients that they should not use Reddit to find strategic information about the admissions process. This is because there are not many experts there, and rather, people without deep background speculating on a topic, like that round 1 is less competitive than round 2, or that schools are looking for a specific career goal or type of extracurricular. Unfortunately, AI can spit out answers based on Reddit that may just not be true at all.

I encourage clients to come to me with their AI research, so we can discuss and validate. I recently had a client who was looking at niche career paths within his industry. AI helped lay out some newer pathways that are feasible from an MBA but perhaps not widely known. Then, we discussed the list of options, and how some of the AI response was a bit too narrow in terms of which schools might fit each path.

AI can be a helpful thesaurus or grammar checker. For instance, Grammarly is a helpful tool to check grammar and suggest adjustments. Sometimes, we find ourselves drafting a resume bullet for instance, and needing a synonym. We can ask Claude what it might suggest.

On the other hand, where we need to hear your voice, we avoid AI. For instance, if you try to draft an essay with ChatGPT (please don't!), you will likely end up with a nicely structured story with polished language and grammar, yet devoid of much reflection or emotion, which is the most important part of this process. It will be very difficult to add your voice in later and fix an already shallow essay. I have unfortunately had to tell a few clients that their essays sounded robotic and cold, which is really not a nice position to be in. I have found it much more effective to draft naturally, in your voice, however unpolished the draft may be, and then use AI later if you need help with a synonym for a word you have used too much, or to suggest ways to make a run-on sentence a bit more concise. Never paste sentences into your essay. Simply ask for ideas, and then use your judgment on how to implement those ideas.

I even find that AI in recommender packets is a mistake, as I would rather your recommenders hear your imperfect voice in your notes to them than Claude's perfect one. The same goes for our client kickoff questionnaire, as it is very hard for me to get to know you when you were not the one writing. I am not judging your grammar, I promise! Even the short answer fields on applications look like they are merely about execution, but they still require strategy. A strong short answer carries a bit of reflection rather than summarizing your resume, and that reflection has to come from you.

Quite a few schools now have a video component to their applications to supplement written content and bring an applicant to life, and I believe part of the reason is that some applicants leaned too much on AI in their written materials for adcoms to glean much. I have heard about applicants who tried to use AI on their video components, receiving the question, and popping it into AI quickly, and reading their answers back to the camera. This is the worst possible thing someone can do on an exercise like this, as it is certainly obvious that someone is reading a script instead of thinking, and it not only shows poor judgment but also lack of confidence in one's ability to simply think deeply and independently.

As someone who looks at admissions materials all day every day, I assure you that it is abundantly obvious when someone has tried to use AI on their materials. Admissions committees can surely see when AI has been used improperly. I would undoubtedly rather see the imperfect version of you in your written or verbal communication than to see Claude or ChatGPT, and I believe the admissions committees would agree.