Anucha Tiemsom/Shutterstock
When OpenAI unveiled ChatGPT on November 30, 2022, the response was modest. Within two months, the platform had amassed 100 million active users, setting a record for the fastest‑growing app in history. This rapid adoption has rippled across every sector that relies on human language, and education is no exception. Universities and schools worldwide found themselves scrambling to determine whether students’ work was genuinely their own or the product of a machine.
For many educators, AI‑detectors seemed like a panacea: a way to distinguish effort from deception. Yet even before ChatGPT existed, plagiarism software had already struggled with false positives, penalizing students unfairly. The new AI tools have only amplified that problem. Students can now produce polished essays with a few keystrokes, offloading the critical thinking that learning is meant to nurture. In the race to keep up, teachers increasingly rely on detection tools that misclassify original work as machine‑generated, leading to unwarranted accusations of cheating.
In this article we examine how AI has impacted schools globally—boosting cheating, fostering tech dependence, and spreading misinformation—while also revealing a path toward more human‑centric education in an era of unprecedented technological change.
Xavier Lorenzo/Shutterstock
ChatGPT and similar platforms provide students with a new shortcut to academic dishonesty. Beyond generating essays, these tools can solve complex math problems (with varying accuracy) and even produce computer code from minimal prompts. Detecting such work is notoriously difficult. A study published in PLOS One had researchers submit fully AI‑generated essays into a UK psychology program’s exam system; 94 % of the submissions went undetected. In computer‑science courses, educators fear that GitHub Copilot and other code‑generation tools will require a complete redesign of curricula.
Incidents of AI‑fueled cheating have surged. Scotland reported hundreds of cases in the past two years, and a Turkish university recently arrested a student who used a camera disguised as a shirt button—connected to AI via a router hidden in a shoe—to receive answers during an entrance exam.
Luis Alvarez/Getty Images
While AI can streamline classroom operations, research in *Computers & Education* shows that students who rely heavily on these tools exhibit reduced agency, learning more by copying than by engaging with content. Teachers, too, are increasingly dependent on AI‑based plagiarism detectors, despite evidence of their high false‑positive rates. A survey by the Center for Democracy & Technology found that two‑thirds of a nationally representative sample of 460 U.S. instructors use such tools.
OpenAI has developed a watermarking technique that reliably identifies AI text, but commercial release is limited by market pressures. Google’s recent efforts may offer a more accessible detection solution. Moreover, some educators use AI to grade essays, sparking ethical debates and raising privacy concerns when student work is uploaded to these systems without consent.
Rob Dobi/Getty Images
Large language models are prone to “hallucinations”—presenting fabricated information in a convincing, authoritative tone. This is especially problematic in education, where the goal is to teach fact‑based reasoning. In December 2023, a student at Hingham High School in Massachusetts received a failing grade on an AP U.S. History project after submitting AI‑generated text that cited nonexistent sources. The school’s lack of an anti‑AI policy led the student’s parents to sue for a grade reversal, a request the federal court denied.
Scientific Reports reports that 55 % of literature reviews produced by ChatGPT‑3.5 and 18 % by ChatGPT‑4 contained fabricated citations. As teachers incorporate AI into lesson planning, the risk of inadvertently disseminating false information to students grows.
Tarikvision/Getty Images
AI systems trained on vast, unfiltered datasets can perpetuate and magnify existing biases. OpenAI acknowledges that its models tend toward Western viewpoints and favor English‑language input. This bias can disadvantage non‑native English speakers, with AI detectors more frequently flagging their work as machine‑generated.
Predictive analytics used in schools—such as tools that forecast high‑school graduation likelihood—often misclassify minority students. A Wisconsin system, in use since 2012, repeatedly identified Black and Hispanic students as at risk, but the algorithm was wrong almost three‑quarters of the time, skewing educators’ perceptions.
Eyesfoto/Getty Images
Despite its challenges, AI can enhance the educational experience when applied thoughtfully. Studies show that AI tools serve as effective brainstorming partners, sparking curiosity and inquiry. Students should first practice original thinking before leveraging AI as a tutor for deeper exploration. AI‑driven feedback and real‑time content generation can create a more responsive, personalized learning environment.
Balance is the key to unlocking AI’s benefits while mitigating its harms. For those interested in AI’s broader societal impact, see our analysis of Elon Musk’s concerns about Google’s DeepMind and the incident involving Google Gemini’s dialogue with a student.
