A survey published March 18 by Kaplan, an education provider that provides resources including test prep and professional certification, received more than 300 responses from pre-law students about their thoughts on artificial intelligence usage in law school admissions essays. The survey focused on four areas: whether applicants should use AI in admissions essays, transparency around schools using AI to evaluate applications, comfort levels with applying to schools that do not use AI in their admissions processes and potential AI bias.
One key finding showed 89% of respondents believed law schools should disclose if they utilize AI to evaluate applications, while 80% of respondents thought AI usage could unintentionally perpetuate bias.
In an interview with The Michigan Daily, Matthew Bui, assistant professor at the School of Information and Digital Studies Institute, explained how AI bias in large language models affects public perception of biases in the context of his course, SI 410: Ethics and Information Technology.
“In the ethics course that I teach, bias is everywhere,” Bui said. “In different fields, bias might mean some different things. But really, at the end of the day, what bias calls attention to is discrepancies or slight skews. And so if data is skewed and then input it into these models and used to train them, then those biases might get reproduced and or even amplified as they’re perpetuated through more and more use more broadly.”
In an email to The Daily, Carson Byrd, associate research scientist in the Center for the Study of Higher & Postsecondary Education, mentioned how the biases in the data used to train large language models can impact applicants, especially as some colleges begin using AI to evaluate application essays.
“While this may seem like a way to efficiently evaluate some application materials with the large number of applicants applying to universities like UNC, Michigan, Harvard, Stanford and other highly selective universities, an underlying issue is how biased AI tools may be for judging the ‘quality’ of essays that can reinforce racialized, socioeconomic and gendered inequalities in admissions outcomes,” Byrd wrote.
Sarah Zearfoss, senior assistant dean at the University of Michigan Law School, said in an interview with The Daily U-M Law School does not attempt to detect AI in admissions essays, instead relying on applicants to sign a certification confirming they have not used AI.
“We don’t attempt to detect it,” Zearfoss said. “We do ask, as with everything in the application … that you sign a certification pledging that you have not used AI to draft your essay among other requirements that everything is your own work. So we are going on faith as it’s a profession. It is a profession where your word is your bond. So that is how we approach all aspects of the application.”
Amit Schlesinger, Kaplan executive director of legal and government programs, said in an interview with The Daily the admissions essay is important for applicants to set them apart, even if it’s not as crucial as the Law School Admission Test.
“The LSAT score is certainly more important, but an admissions essay, if written poorly, can harm a student’s application,” Schlesinger said. “On the other side of that, it gives students and applicants the opportunity to share a unique story, be authentic and sort of highlight from a personal experience something that at least sets them apart from other applicants.”
Second-year Law student Kristia Postema said in an interview with The Daily applicants can manipulate AI to write admission essays in a way that makes it still appear human-written.
“I think it depends on how you prompt the AI,” Postema said. “A lot of times, they’ll ask AI to write something in a specific tone, or they’ll explain what the essay they’re asking AI to write is for, and so I think that that makes it sound a little more genuine, because then it is able to like try and add like emotion into it.”
But as AI gains popularity and policies remain vague, Schlesinger emphasized the importance of reaching out to the admissions office of law schools if uncertain.
“I wouldn’t say that admission policies will be overwhelmingly against the use of AI,” Schlesinger said. “But certainly they should and will have a policy that is transparent for the students, and I think that’ll be critical for the application process and our current recommendation for students who don’t see a clear policy is to reach out to the admissions office of that specific school and ask, because knowing whether or not it’s permissible or not is certainly beneficial.”
Zearfoss says that while the University doesn’t allow the usage of AI, the policy could change.
“I think it’s quite possible that at some point in the future, we would change our rules and our guidance,” Zearfoss said. “I evaluate them every year, but not right now.”
Daily Staff Reporter Kayla Lugo can be reached at klugo@umich.edu.