Illustration of an open laptop with an acceptance letter to the University of Michigan on the screen, with maize and blue ribbons coming down
Design by Iris Ding.

“Congratulations! I am pleased to inform you that you have been admitted to the University of Michigan!” is a phrase any high school student applying to the University of Michigan hopes to see come decision day. In the past few months, however, new technologies like the text-generating software ChatGPT have gained popularity and now pose a monumental threat to universities’ ability to evaluate candidates. These technologies make it easy for applicants to submit fraudulent work — all while reliable AI detection technology may still be years away.

If the University does not take action to ensure that AI programs don’t negatively affect the integrity of the application process, applicants who would normally be admitted are at risk of being deprived of a chance to thrive. Although there has been progress in software that could possibly detect when AI is used to write essays, the University should also move toward adding interviews conducted by its large alumni base in order to offset the possible issues that could arise in the future with ChatGPT.

Currently, ChatGPT sees more than 13 million daily users per day. Some of the more notable tasks this AI is used for include assisting with coding, passing a law school exam and writing long-form academic essays. While these tasks have shown the impressive ability of the new AI, ChatGPT’s ability to write essays raises a point of contention for the U-M admissions process.

While essays are not the only part of the application process, they are one of the main criteria evaluated in an applicant’s application. ChatGPT, whose essay-writing abilities have come close to deceiving college professors, poses a substantial threat to the admissions process as applicants could use it to write their essays — potentially giving them an unfair upper hand. This issue will only continue to be exacerbated by the rapid growth of AI’s capabilities. AI-generated text could create unfair advantages for applicants through its ability to summarize information and make strong arguments.

The University has shown recognition, but not action, thus far when dealing with ChatGPT. In an email interview with The Michigan Daily, University spokesperson Kim Broekhuizen told The Daily that there are not measures in place to combat the use of AI in applications.

“At this time, our admissions process hasn’t changed related to ChatGPT,” Broekhuizen said. “(The University is) aware of the new technology and are monitoring and collecting information.” 

Because ChatGPT’s writing abilities could threaten the University’s admissions process in the near future, there are some specific changes that should be made to offset the impacts of text-generating AI before it is too late. One of the changes that should be made is the adoption of the new AI classifier being developed by OpenAI. This program would be able to detect the use of AI in essays and be able to identify them for admissions officers. However, this system is not fully reliable yet, as it is still being developed and is not 100% accurate.

Although the classification software would likely be able to detect the essays that were written by AI once it is fully developed, there is still a possibility that essay-writing programs or other auxiliary programs will evolve to bypass detectors. This risk can best be seen through the software and methods already developed to bypass plagiarism-detection softwares currently used by academic institutions. Because of the shortcomings of these softwares, the University should also consider conducting interviews as part of the application process alongside its required essays.

The University should take this potential change of admissions criteria into serious consideration as interviews AI software could not tamper with them. This addition would make the admissions process more accurate, as it would better ensure that applicants are who they claim to be on paper, while also serving as a way to let applicants better express themselves. 

Although the integration of the interview into the application process would require the admissions office to change the main criteria for judging an applicant, the added benefit that it would give to the process would outweigh the disadvantages. 

The disadvantage of adding an interview as a part of the application process would be its integration. The admissions office would be required to change their main criteria to include interviews, which would cause a greater reevaluation of the admissions process as a whole. The implementation would be difficult as well. Other schools that include interviews are typically smaller in comparison to the University, so it would require a lot of work to fully implement interviews on such a large scale. 

On the other hand, the advantages of adding interviews can be shown in how applicants are evaluated. Through interviews, the University is able to better understand their applicants beyond what is written down and another opportunity to better express themselves in a form that may be more natural or beneficial for them.

Notably, Georgetown and Northwestern conduct their interviews using a pool of alumni, conducting interviews in their local region instead of at the university itself. If the University of Michigan would call upon its over 600,000 alumni — one of the biggest networks in the world — the institution would be able to have enough interviewers per region without having to overwhelm its alumni population.

Overall, interviews give the University the ability to vet the applicant beyond the essay, which could be tainted by the use of AI-generated text — the ability to learn about the applicant without the risk of AI interference. 

The admissions process is at risk in the near future as ChatGPT and other programs continue to improve and become more widespread in their use. As applicants possess an increasing capability to bypass writing essays, the U-M admissions office needs to adapt to ensure that the process is not subverted. 

Tom Muha writes about issues relating to social media & the internet. He is an Opinion Columnist, and can be reached by email at