Is AI Better at Evaluating College Essays Fairly Than Humans?

0
313


typing on a laptop

Getty Images/skaman306

Every year, prospective college students are challenged with writing an admissions essay that not only tells their story but also makes them stand out in a way that compels the admissions officer to accept them. A new AI tool could transform that process.

Researchers at the University of Colorado Boulder have developed AI tools that can read and evaluate college admission essays, according to a report from Daily Camera.

Also: Generative AI can be the academic assistant an underserved student needs

Partnering with the Common App, the researchers assembled a data set of 300,000 students who applied to college in 2008 and 2009, gathering a random sample of essays that were read by humans and rated on personal qualities such as leadership, mindset, and teamwork.

The AI tool was modeled to do the same, and the research team found that the AI-generated ratings matched closely with the human ratings. According to the report, the AI’s ratings could also predict fairly accurately whether or not a student would graduate.

The tool is meant to help students who might otherwise get overlooked by highlighting qualities within the essay that may make them great candidates for admission, even if the essay itself wasn’t perfectly written.

These tools have another advantage: They can circumvent the personal bias that an admissions office could hold regarding an applicant’s race, gender, or socioeconomic background.

“There’s so much more to a person than what they write in an essay,” said Sidney D’Mello, CU Boulder professor and the study co-author, to Daily Camera. “I really think these tools can help surface aspirational, deserving, amazing students who otherwise get overlooked.”

Also: How ChatGPT (and other AI chatbots) can help you write an essay

Even though the AI tool can give unbiased feedback, it is not meant to replace human admissions officers.

Instead, D’Mello shared that the tool is meant to supplement the admissions officers’ inferences, highlighting the applicants’ character traits from their essays that the officer may have overlooked in search of an excellently written essay.

As with any AI model, there are present risks, and the researchers reassured the public that before the tools are used, additional work is needed to ensure accuracy and efficiency.