Why a Student Might Use AI, False Positives & How Colleges are Responding
🤖
Prompt Coach LMS Database/🤖Artificial Intelligence/🤖Why a Student Might Use AI, False Positives & How Colleges are Responding

Why a Student Might Use AI, False Positives & How Colleges are Responding

 

Why Students Use AI

We recognize that dealing with AI can be a frustrating aspect of our work as educators and Writing Coaches. We hope that by considering the circumstances and motivations that might lead a student to use AI, you will feel empowered to be proactive and collaborative in your approach.

Resource Access

One of the main reasons students turn to AI for help on their college admissions essay is because they don’t have access to other sources of guidance or support. In these instances AI can be a useful tool.
Because the students we work with do have access to at least one excellent resource (you), we generally discourage the use of AI during any part of the writing process. However, we recognize that students submitting their essays to us with zero other support may find the temptation to use AI even greater.
✴️
If a student only gets one review with us, they may think using AI to get a “stronger” first draft is a smart way to get the biggest return on their investment.

Brainstorming Without a Plan or Personalized Effort

Some students have teachers encouraging them to use AI as a brainstorming tool. While AI can be extremely valuable for brainstorming, its success really depends on the training a student has received in crafting prompts for the AI.
  • If a student is just plugging in the prompt plus a few details, they are going to get a generic and buzzword-driven response that doesn’t serve them and encourages them to write in hyperbole.
  • If a student writes a more comprehensive paragraph about the experiences they want to use in their essay and their theme, AI could generate questions to help them brainstorm unique entry points. With this process, they could end up with a compelling first draft! Still, they should answer those questions themselves.

Let AI Do It

Of course, some students think that AI could do a better, faster job than they could. We have seen this from students who had consistently been “doing the work” but then found themselves with half a dozen essays to write in just a few days, so they panic and turn to AI. Other students may have successfully used AI for assignments in the past or never received feedback on their written work.
Sometimes, if we can figure out a student’s motivation for using AI content generation, we can support them in reverse engineering a human solution.
  • For example, if a student is worried about the level of their writing, we can assure them that stylistic and technical editing should come after we work on developing content.
  • If a student isn’t sure where to start or how to organize their thoughts and “AI does it better,” we can (1) point out the holes in AI’s work and (2) help them develop an outline to organize their ideas.
  • If a student feels overwhelmed by a block of brainstorming questions, we can break them down into in-text brainstorming questions that identify key entry points for insights.
  • If a student just thinks there is too much work to do. They’re right! College applications are a lot of work! We can use content mapping and derivative essays to work efficiently instead of wasting time on multiple drafts to correct superficial and non-congruent AI-generated content.

False Positives

In recent years we’ve seen an increase in false positives among native English speakers. Counselors, parents, and even Writing Coaches witness a student composing their essay, and still, an AI detector or another editor flags the submission. We also hear frustrated and devastated responses from students who feel their genuine hard work is being dismissed with a false accusation. Because these are conversations you may be a part of, it’s important you appreciate where these false positives are coming from and how this situation has affirmed our foundational approach.

Increased Exposure

Students are exposed to AI writing often, early, and frequently without knowing that AI was the author. They might see AI-style writing celebrated or rewarded in the classroom or online. It’s therefore not surprising that when students go to write themselves, they (intentionally or not) mimic AI’s signature style and voice.

High School-Level Writing

High school writers, especially in early drafts, will use generalized and broad language, take meandering detours in their narratives and their tone, and rely too heavily on buzzwords instead of focusing on personal, unique insights. And, many times, a first draft conclusion will have nothing to do with the rest of the essay! Students often don’t know how to get to where they think they want to go!
Unfortunately for those of us who work with high school writers, those are the same issues often present in AI-generated and AI-polished essays. These correlations make identifying AI in our student population particularly tricky!

Polishing—and Grammarly’s Overreach

Proofreading services like Grammarly often obfuscate exactly how much of a student’s content they are changing. A student might click through or hit “accept all”, thinking they are only accepting minor edits to polish grammar and spelling, but are actually accepting modifications related to tone, flow, word choice, or other subjective aspects of writing.
Letting Grammarly Premium have a go might make their essay sound “better” superficially, but that is often at the cost of personality and unique insights/details.
  • Suggestions that are underlined in red are related to grammar and spelling. That still doesn’t mean they should be accepted without review! AI sometimes (often) gets the context wrong! Suggestions in blue are stylistic suggestions, and these should be accepted with caution and sparingly. Leaning into AI’s suggestions for voice and tone can quickly homogenize your feedback and your students’ work.
  • While we require that coaches use grammar & spell-checking software, we strongly recommend caution among high schoolers who are less capable of determining what is a suggestion and what is a correction. We’ve also found Grammarly to be particularly guilty of overreach and encourage you to consider a different service, such as Quillbot or Hemingway.
    • If you do use Grammarly, there are options in account settings (Account > Writing > Your Preferences) to turn off some suggestions. We recommend turning off the options that start with “sound more…” (confident, personal, etc.) and “rewrite text…” (for clarity, improved effect, etc.). These seem to be the suggestions that overstep the most and veer into AI territory.

How Colleges are Responding

AI has a huge impact on writing style, even if students just use it to polish their writing. As such, colleges are looking for authenticity and personality in essays more than ever!
Essays are very much part of our understanding of the applicant, we’re just no longer assuming that the essay is an accurate reflection of the student’s actual writing ability. [Essays will now be used to] help understand the applicant as an individual rather, not just as a set of attributes and accomplishments. [The admissions office now values essays that give] insight into who the unique person is whose application we’re reading… content and insight matter more than style.
Dean of Undergraduate Admissions Christoph Guttentag, Duke University
Some colleges are employing AI checkers as part of their review process, while others are simply discarding any application with suspected content. (Exact stats are tricky, as many schools aren’t forthcoming on their process and/or prefer phrasing like “we use a holistic process” and “we use software only for data analysis.”) Because the process of confirming AI can be very time-consuming—and because the admissions landscape is so competitive—even if a single section of an essay is flagged as being potentially AI-influenced, it’s likely the entire application will be rejected.
That is why, regardless of why or if a student used AI, it’s important for us to understand what might flag as AI and have strategies to address those issues, which is exactly what we’ll cover next!