There is a significant amount of concern expressed about AI generated essays. The main reason I would be concerned about it has nothing to do with the graduate admissions process: Becoming dependent on AI makes people stupid. MIT's study on this issue is worth reviewing. See abstract (https://www.media.mit.edu/publications/your-brain-on-chatgpt/) and an interview with the primary researcher (https://www.media.mit.edu/articles/a-i-s-effects-on-the-brain/). I would also recommend INSEAD Dean Mortensen's Have You Really Counted the Costs of GenAI? as it explains in detail what is lost by over reliance on AI specifically in reference to the cognitive skills required for success both academically and professionally. Note, my use of the word "dependent." I am no Luddite. I think AI is useful if used effectively. That is why I am not against using AI for making MBA or graduate applications. It has its place.
For MBA Admissions purposes in particular, my AIGAC colleague, Petia Whitmore, has put together an incredibly helpful article for all MBA applicants on how to use and not use AI. I will not repeat what she has written there.
Instead I want to make a different point: THERE HAVE BEEN AND ALWAYS WILL BE SHITTY ESSAYS AND AI IS JUST A VERY EASY WAY TO CREATE THEM!
I like the word "shitty" because it is captures the full essence of a bad essay: It stinks. Humans have been writing shitty essays forever. Hence AI, utilizing the corpus of content upon which it provides answers, will produce some shitty essays. Based on what I see it can certainly do that. Again, not all essays generated by an AI will be shitty, but if you don't have a human with the actual capability to judge what constitutes a non-shitty essay, problems will arise.
Some thoughts on the continued role of human incompetence in the essay writing process: What makes an essay shitty?
It fails to answer the question. People have been writing essays that did not answer the question correctly forever. Even with AI, I receive shitty essays from clients that totally fail to answer the question because (1) the writer did not use the correct prompts, (2) the writer did not insert the correct raw data for the essay, (3) the AI failed to really address the question being asked, and/or (4) the writer did not understand the question and therefore could not effectively judge the AI output. Only a person with good judgement can determine whether an answer is effective or not. Many applicants can certainly do that themselves, but not everyone.
It lacks strategy. Since MBA and other graduate admissions essays have only one sole objective, which is admission, you would think getting the strategy right would be easy. But this is the hardest part because what will work for one candidate will not work for another. That is often why advice from admits fails to help an applicant. Prior applicants (alumni and current students) tend to give advice based on their own experience. However, what got Jane admitted is not the same thing that will get Mary admitted. They are different people with different strengths and applying at different times. What worked at one point in time simply no longer always works at a different point in time. AI can provide strategic options and help with brainstorming but it can't provide holistic judgement that fully accounts for a particular candidate. Candidates and good advisors (professional or otherwise) must make the ultimate judgment about what will work best in an application now.
It is boring. I have been an admissions consultant since 2001. The number of absolutely dull essays I have read and then helped a client make engaging is something I have never counted, but well into the thousands. How to make a shitty boring essay: Tell an obvious story that contains no real development, lacks detail, does not attempt to connect to the reader on an emotional/intellectual/mission level, but is perfectly rational in the most soulless way possible. Often they are mere extended versions of a resume bullet point. Many people don't know how to tell a story, especially in writing. It is not a skill distributed evenly throughout the human population, which is why we value those who do it well. That said, story telling can be taught. I know because I have taught it to thousands of people over the last 24 years. However, AI's are not inherently good at storytelling unless you provide a prompt with such sufficient details that you might have well written it yourself. They will tell a story for sure, but not inherently a good one. If you don't know how to tell a story, chances are unlikely you will be inherently good at knowing whether an AI produced story is any good. I can say this because I have 24-year sample size that makes it perfectly clear to me that many people can't tell a good story about themselves without being taught how to do it. A language engine, an AI, is not an aesthetic judgment engine and also not a judgmental reader, which is what required for judging a story.
It lacks sufficient details. This is a subset of boring to a certain extent but worthy of its own categorization here. For as long as I have been an admissions consultant, I have been impressed by the ability of some writers to be completely blind to the importance of detail in MBA essays. It is though they never read a news article, story story, or any narrative prose. No time provided. No company name. No characters in stories (such as teammates in an essay where the theme relates to teamwork). No details about how the applicant solved a problem, succeeded, failed, etc. No numbers to back up their arguments. An AI will certainly populate a story with detail, but that requires providing the AI with the detail or adding it later, which are both dependent on the human knowing they need to have such details.
It is based on bad argumentation, often found in the conclusion. A good example of this is an essay that discusses learning something when, in fact, what was supposedly learned, was clearly already known by the applicant because the actions that applicant took in the situation being described demonstrate that. For example, an applicant writes about how he or she collaborated with a team by using their existing teamwork skills and then writes about how they learned the power of collaboration from this experience. It makes no sense, since they already knew how to collaborate, so they did not learn that. They already knew it. False learning is any situation when you indicate that you learned something, but actually it was something that you already knew or others are likely to assume that you know. False learning tends to undermine the credibility of applicant in terms of their intelligence and honesty. A related form of this, which AI argumentation seems to produce is that an experience is reaffirming what the writer knew already. Reaffirming something is not learning something either. Again I have read thousands of essays that suffered from bad argumentation. Why wouldn't AI fix this problem? Because the world is filled with bad arguments and an AI can't always distinguish between a good and bad argument as both can be found in the corpus upon which it makes judgements.
Big Takeaway: Anyone can use AI, just like anyone can use a keyboard, but the result is still dependent on human judgment. If someone's human judgment skills are undermined as a result of an over-reliance on having an AI do their thinking, it is best that they never get into any kind of position of responsibility where they would be required to think because they will be unable to do through lack of practice. In this sense, AI rather than enabling the incompetent and lazy to gain entry into MBA programs will likely have the opposite effect if the reader is a discerning human being (I will assume that is the case with most admissions officers). The opposite is also true, those that are competent and hardworking and use AI effectively are likely to have a significant advantage in the admissions process and in their professional lives. At least is that is what I hope for, otherwise full corporate Idiocracy is likely to be achieved within my lifetime.
