A recent CNET article, reporter Rachael Kane discusses how to recognize signs of AI-generated writing and why it can be difficult to detect with certainty. She notes that AI-written text has identifiable characteristics beyond punctuation and phrasing.
She highlights that AI tools like ChatGPT are increasingly being used to generate low-quality content, raising concerns about their impact on education and workplaces. Kane, drawing from her experience as a professor, describes AI as the ultimate “work smarter, not harder” shortcut that is now visible in classrooms and professional settings. While these tools can be useful for brainstorming and drafting, she notes that they can also produce sloppy or low-effort work when used without care.
From her own teaching experience, she adds that tools like ChatGPT and Claude are becoming easier to identify. She points out that AI-generated writing is often grammatically correct but predictable and lacking personality, sometimes described as having a “Wikipedia Voice.” She explains that it can rely on clichés and avoid human-like imperfections, which can make the writing feel artificial or overly polished.
Professor Kane explains how to identify AI-written content. She notes that AI-generated writing often repeats key terms, may include inaccuracies, uses unnatural sentence structures, and lacks a natural human tone. She also points out that students sometimes use tools like ChatGPT to generate essay responses.
Kane demonstrates this with a ChatGPT prompt such as: “In 300 words or fewer, explain how this SWAT and brand audit will inform your final pitch.” She explains that she has received very similar responses from students, often repeating prompt language and key terms. She says this pattern can indicate AI use and that she has developed strategies to detect it in academic work.
Below are some of Professor Kane’s approaches for identifying AI-assisted writing in her classroom:
- Understand AI capabilities: Kane explains that modern AI tools can generate complete, structured, and even cited assignments in seconds. She recommends becoming familiar with tools like GPTZero and Smodin, which are designed to detect AI-generated content and help educators better understand how such tools are used.
- Do as the cheaters do: Kane suggests that educators try using ChatGPT themselves to complete assignments. This helps them understand the quality and style of AI-generated responses, making it easier to recognize similar patterns in student submissions.
- Get a real sample of writing: She advises collecting a baseline writing sample at the start of the semester. For example, asking students to write a short personal reflection such as “200 words on your favorite toy as a child” allows instructors to compare natural writing style against later
- Ask for a rewrite: If AI use is suspected, Kane suggests asking students to have AI rewrite the same assignment. She notes that AI tools often make surface-level edits, such as substituting synonyms without changing structure or meaning. In her experience, even when she tested her own writing, the AI made it more polished but also more generic.
- Can you always tell if AI wrote something? Kane emphasizes that strong evidence is essential when making judgments. She encourages maintaining healthy skepticism during grading while also developing a solid understanding of AI tools to support fair evaluation.
In summary, Kane encourages educators to engage with AI rather than ignore it. She suggests focusing on making learning more meaningful and engaging than the act of cheating, and avoiding frustration when students use AI tools.
Acknowledgement: This summary was generated using AI writing tools as part of my AI Explorations process. For more information, please refer to the original article listed below.