AI Use Detection in Application Essays

It is a challenge to write even one compelling application essay that can impress the committee; what if you have to compose several at once? Let’s go further. What if you are a counselor and you need to write dozens of recommendation letters a month and make them look convincing and of course different? They say, one in three college applicants write their essays with the help of AI tools, so does that mean that college application process now makes no sense as it is no longer fair? Isn’t it high time for us to actually introduce AI detection in college essays at every stage, including that of submitting papers to apply? Have any of the colleges started this procedure and what are the consequences of neglecting this check? There are a lot of questions to be raised and their number will definitely keep growing in the current era of AI.

Academic integrity has never been under such a severe threat as it is now. The students and even applicants feel empowered with the fast adoption of ChatGPT, GPT-4, and other generative AI tools. A risk factor is that they do not view it as academic integrity breach! So, if they do not copy and paste the texts generated by AI tools, they believe that it is academically honest and ethical to make use of this assistance. With the use of correct prompt patterns, academic writing with the help of GPT-4, for example, can guarantee first-rate results. What is more, its proficiency keeps enhancing and there are no more issues with critical analysis, research, or presenting complicated ideas in a simple way as the tool does everything by itself. There are more and more ChatGPT admissions in colleges as excellent language capabilities of the tools are used not only for actual writing, but also for:

• generation of ideas;
• brainstorming;
• effective structuring of the text;
• creating an outline;
• composing a draft;
• doing review, proofreading, and editing.

No more writer’s blocks as the applicant can get hundreds of tips to use in no time. Even a weak essay may be effectively refined quickly to get to the highest level of integrity and clarity of presenting ideas. Besides, what can be a more convenient starting point for writing than mere providing prompts to an AI tool?

A research paper requires data analysis and profound exploration of ideas and facts. If it goes about a literature review, it deals with looking for a new perspective at a certain topic. A comparative essay demands reasoning based on evidence. Lab reports on real scenarios check on the scope of scientific knowledge and analysis of experimental findings. An application essay is associated with the ability to impress the readers and stand out among many others of the similar kind. Whatever type of writing you need, you may rely on the AI assistance if you have provided proper prompts and instructions. Numerous questions related to academic integrity AI still need answers and solutions.

Tailoring admission essays has always been an art that can open doors to a new life. Non-authentic writing will never impress readers as the story will never be believable and likable without your own voice in every line. Natural flow and compelling ideas mixed with not forced style can show the commission who you are and how you can benefit the institution. It is a paradox that you have both to sound natural and be driven with a purpose. You not only recollect situations from your own life, but use them as illustrations to your talents, values, and abilities. Miles Davis once said, “Man, sometimes it takes you a long time to sound like yourself.” That is true! It is not easy to develop a strategy of writing an application essay that will sound authentic. The question is whether artificial intelligence can generate authentic texts.

This article will focus on current practices, technologies of AI detection, their efficacy, and ethical implications of using AI tools in writing application essays.

How AI Tools Have Burst Into Modern Life

Since November 2022, ChatGPT has been making humanity amazed with its explosive growth.

• It took only 5 days for it to get over 1,000,000 users! For Netflix, it took more than three years.
• Monthly, the number of visits to ChatGPT exceeds 4.79 billion!
• YouTube is the main channel for social media traffic of ChatGPT.

The world was taken by an unexpected storm! Wikipedia refers to ChatGPT as the “fastest-growing consumer software application” as in only about two months, the number of users exceeded 100 million. The adoption of AI tools has been accelerated with COVID-19 as well, and education has found itself in the era of digital pedagogy. All news and social media posts had ChatGPT as one of the hottest topics for discussion, while educators had rather mixed feelings about the extraordinary abilities of AI. Is it a threat for schools or the future of learning and teaching? That is the main dilemma teachers faced and they are still trying to find the answer to the question.

However, for the students, AI has opened all the doors that used to be shut. Revolutionizing the educational system, ChatGPT at once let the students forget about all the challenges of brainstorming, editing, and even drafting essays. It became an Internet sensation that gained enormous popularity among the applicants, students, and bloggers. Since then, the number of searches for ChatGPT and other AI tools has been growing exponentially.

To Use or Not to Use? That Is the Question

Your passionate dream since your childhood years has been to become a student of the university from which your dad graduated. You know that the competitiveness rate of the applications to this university is unbelievably high. Will you say No to a chance to get assistance in writing an application essay and try to get some head start? We sincerely doubt that.

So, you need a perfect application essay for the admission committee to see you among others. You have to make them interested in your personality, background, talents, and integrity. It is clear for you and your family that you deserve a great future, but our task is to convince the people you do not know that you are going to be a great student. The only way to give information about yourself to them is through your writing.

“Did AI kill the college admission essay?” This heading of an EdScoop article wants us to get an insight into the problem that is getting more and more burning nowadays. A college application AI has evolved a lot, but it still shows the advantages of a certain applicant who wants to show that they are prepared to be successful in their studies. It is focused on showing personal values, beneficial traits, and sense of purpose.

Is Fairness Still Possible in Admissions?

With the onset of artificial intelligence, one of the greatest academic concerns was that the students can submit AI-generated application essays and win more points not in a fair way. Forbes refers to the words of empowering guidance at Georgia Tech, given to the students on the use of AI: “Tools like ChatGPT, Bard, and other AI-based assistance programs are powerful and valuable tools. We believe there is a place for them in helping you generate ideas, but your ultimate submission should be your own … take the same approach you would when collaborating with people. We think AI could be a helpful collaborator...”

Still, the review process in numerous institutions is also done with the help of AI! The survey of the Intelligent showed that about 50% of higher education admission offices were reviewing applications with the help of AI in 2023 and the number was going to rise to 80% the next year (Forbes). Scanning through the essays, the tools pick out the proofs of personality features that colleges need, like leadership, diligence, or persistence.

While some of the schools implement AI-detection protocols to identify AI plagiarism, others remain uncertain or silent on the issue. While some review applications with the help of AI tools, others tend to use the outdated traditional methods. Still, the overall picture is changing all the time.

Policy and Practice Patchwork in AI College Detection

‘If we're unified, there's nothing we cannot do,’ said Ray Nagin. Without a unified approach, it is unlikely to reach good results in maintaining academic integrity. Kelly Dore in her article for Campus Technology claims that AI ‘has officially taken root’ in the sphere of education. While some of the professors recommend using AI to complete certain assignments, others ban its use completely and threaten with expulsion. At the same time, certain instructors and tutors compose feedback texts for the students with the use of AI. Leveraging the use of AI tools in an ethical and responsible manner is a must nowadays.

Campus Technology’s report cites stunning statistics that should stimulate all the involved parties to start taking immediate actions. In March 2024, over 6 million out of 200 million reviewed papers were almost totally generated by AI, and over 22 million were partially generated. PlagiarismSearch AI Content Detector, GPTZero, and Turnitin AI detection are only some of the popular AI detection tools that are supposed to fight the threats. Their advantages are clear, but still there is always only a likelihood rate of using AI, not the absolutely undoubted result. Potential false positive rate (for instance 1% in Turnitin) still creates serious limitations in AI text detection.

Intuition vs. Technologies: What Do You Vote For?

It is a great challenge to strike the right balance between the use of human oversight and logical thinking capabilities of AI. Some admissions officers and teachers still try to detect all the inconsistencies in the texts to outfox the AI algorithms and be fair in judging the applicants.

Jonathan Gillham, a Forbes Councils Member, presents the results of a survey which state that teachers are mostly confident in their ability to detect AI text. Humans often think they can identify AI-generated copy on their own. One study found that teachers had a high level of confidence (77.3%) in their ability to identify AI text in student assignments and differentiate between human and AI-generated ideas (Forbes Tech Council).

Still, the data states differently:

• The accuracy of human identification of AI-generated text is approximately 45%.
• With advancements in AI tools, the rate of accuracy of human detection gets lower (57% with GPT2 to 49% with GPT3).
• Students can identify AI content in 60% of cases.
• The level of misidentification in truly authentic scientific abstracts is high. Mostly teachers identify truly human text as AI-generated in 38% of cases.
• The analysis of residency applications showed that the level of accuracy in identifying AI text is about 65%.

It is absolutely clear that a teacher cannot ensure identification of AI-generated content without the help of special tools. Otherwise, authenticity will be under risk. Only a combination of human reasonability and judgment with the use of AI detection tools can help instructors address the challenges with assessing papers and application essays.

Student Usage Statistics

No one can deny that the students use AI tools in writing whatever the rules are and this practice is going to stay. One in three students always uses AI in writing. It is always interesting to see the statistics of what the students say about that:

• 53% of students claim that they have already had tasks which demanded the use of AI.
• 80% of students claim that the use of AI was discussed in the classroom.
• 58% of students state that there are certain policies on completing tasks with the use of AI in their schools.
• 28% of students indicate discrepancies in the policies of professors in terms of AI use in the same school (BestColleges).

The survey conducted in 2023 by Pure Spectrum showed the following results. Getting back to college application essays, only 56% of students think that using AI is unfair as it gives advantages to some students, and 25% are neutral about that. The opinions of the students on whether it should be an accepted practice for the students to write admission essays with the help of AI tools vary: 37% believe it should be allowed, 26% are neutral, and 37% do not accept the idea (BestColleges).

A scientific study (Nature.com) of Yijun Zhao gave an evaluation of using traditional models and transformer-based LLMs to detect the nature of written content. While general AI content detectors show some limitations, classification models developed for specific domains show their potential. The study also revealed that it is more difficult to identify when the text is revised with AI than when it is generated totally.

Stanford scholars also state that in case the writer is a non-native English speaker, the reliability of AI detectors is rather low. More than 97% of TOEFL essays were claimed to be AI-generated although they were written by non-native English students (Stanford HAI). Thus, a question arises about how objective AI detectors are. If a non-native English speaker submits an application essay, they may be falsely accused of cheating!

Infographic: Student Usage of AI Tools

Note: These numbers are derived from various student surveys, including BestColleges and data from a 2023 Pure Spectrum study. Percentages represent a subset of the relevant student population and may vary depending on survey scope.

Future Trends and Innovations in AI Detection

New approaches to AI detection are underway! For instance, OpenAI has developed a text watermark system and a special detection system for texts generated by ChatGPT and they are going to introduce it really soon. Still, the company profits will be threatened with this new approach as ChatGPT users may lose interest in using it. For sure, there are certain techniques that let the users be smarter than even the most advanced system: they translate the texts from one language to another, add emojis, and do other tricks.

Schools do need more academic research that can make AI detection more effective, but the question is whether it is profitable for AI business. What is needed is the following:

• clear consistent policies about using AI-generated ideas in admissions essays;
• a combination of human judgment and use of sophisticated AI detection tools in checking the written texts.

Although according to a Professor of the university (Business Insider), AI tools are “not the threat — a lagging and outdated approach to education is,” it is time for fighting cheating and praising originality.

Innovation vs. Integrity

There are a number of ethical dilemmas related to AI detection in application essays and other types of writing. Frequent cases of biases and potential false accusations make the use of AI tools more and more challenging for educators. The atmosphere of mistrust in colleges does not serve as a beneficial upgrade in classrooms. A single false flag may cause very serious consequences. A minor error rate of only 1% can lead to having more than 200,000 essays falsely flagged for AI plagiarism a year! The students will lose their scholarships, get penalized, and have ruined future instead of bright opportunities they have been dreaming about. Who will make it up for the anxiety and stress they experience? There is no answer to that.

The role of AI in admission essays is one of the spheres in which there are a lot of open-ended questions.

• What is the way to find out if the applicant submitted an authentic application essay?
• How can the schools enforce new rules for the new students?
• Which way is it possible to prevent getting unfair advantages with the help of AI?
• What is the ethical way to use AI to ensure fairness and maintain integrity in education?

Inconsistent practices in different colleges and institutions along with certain restrictions in the ability of AI plagiarism checkers result in more and more challenges for educators. The problem is actually twofold: on the one hand, it is a must to detect AI-generated essays and on the other hand, to develop new assessment methods which would be adequate for a new level of technological advancement. It is high time that all the researchers, practical educators, and policy-makers started looking for solutions in effective collaboration. Integrity and innovation require the right balance, but it is not easy to find it.

College applicants need clear guidelines and restrictions in the use of AI tools from the very beginning of their academic journey, while the traditional practices of admission should be reviewed and reexamined without any further delay.

Nothing will change if you, educators and admission professionals, do not research and discuss the problem, sharing your best practices. Social media now should become the stage for changes in education! The main topic in academic forums should be AI and its ethical use. Only if the public is aware of the problems and open for introducing important policies, there will be some transformations in the system itself.

References

• Wikipedia: ChatGPT
• Education in the Era of Generative AI (ResearchGate)
• EdScoop
• Forbes: College Admission—An AI Revolution
• Campus Technology: Managing AI in Admissions
• Campus Technology: Turnitin—Use of AI to Write Papers
• Forbes Tech Council: Can Humans Detect AI?
• BestColleges: Students Using AI for College Admissions