GoKawiilListen to this post
Key Takeaways Students are bypassing AI detectors by using tools to make their text sound convincingly human.
AI detectors are unreliable, with high rates of both false positives and false negatives, making it difficult to detect cheating, according to a New York Times report.
At Harvard, professors are reacting to AI by weighing pen-and-paper tests more heavily than other assignments in final grades.
One look at social media makes it clear — it is easier than ever to cheat with AI and get away with it, leaving students with an ethical challenge.
Students are increasingly using AI writing tools to complete essays while slipping past school detection systems, The New York Times recently reported. Short videos on TikTok and YouTube walk students through step-by-step instructions for generating essays with chatbots, then adjusting that text so it appears as though a human being wrote it.
These videos advise students to treat AI as a ghostwriter rather than just copying and pasting from a chatbot. In other words, students run prompts through ChatGPT-style tools, then process the output through additional apps to disguise its origin before submitting it as their own.
A key tactic involves “humanizer” tools that rewrite AI text to look more like something a student would write. These services add variation in sentence length, change up vocabulary and sometimes add mild errors or slang to break the smooth flow of words that many detectors look for.
“Autotyper” tools go further by simulating the physical act of writing. They slowly type the essay into Google Docs or other platforms in real time, complete with pauses, edits and occasional backspacing, leaving behind a version history that looks like a genuine draft rather than a pasted block of text. Some even sprinkle in intentionally misspelled words.
Why AI detectors keep failing
... continue reading