Seven Ways to Spot AI-Written Text
A significant portion of online content now comes from AI chatbots. Text generated by these systems is harder to detect than AI images or videos, but it's not impossible to identify. Look for these seven red flags.
1. Unnecessary Encouragement or Enthusiasm
AI chatbots affirm whatever viewpoint a prompt suggests. When you ask them to evaluate something, they'll praise it-often with enthusiasm that doesn't match the actual context.
Ask ChatGPT about your skincare routine, and you'll get headers like "What You Did Very Well" before it tells you that you "hit the three pillars of skin aging prevention almost perfectly." A human reviewer would offer actual criticism.
2. Overly Perfect Grammar and Punctuation
Real writers break grammatical rules for effect. They use sentence fragments, run-on sentences, and make mistakes. Most people don't have encyclopedic knowledge of grammar rules, so errors slip through.
AI chatbots rarely make these mistakes unless explicitly asked. When asked to write free verse poetry-which permits breaking every grammatical rule-ChatGPT still produces grammatically correct sentences with line breaks. A human poet would experiment much more freely.
If a long Reddit post contains no grammatical issues whatsoever, that's worth questioning.
3. Routine Inclusion of Em Dashes
Em dashes break up sentences for emphasis. Chatbots use them far more frequently than average writers do.
One instance doesn't prove anything. But if a piece contains multiple em dashes throughout, it's a potential indicator.
4. Repetitive Phrases or Words
Certain words and phrases appear repeatedly in AI-generated text: "harness," "illuminate," "pivotal," "at its core," "delve into," "that being said," "to put it simply."
Seeing one of these phrases means nothing. But if multiple appear in a single article or post, that's a signal worth noting.
5. Heavy Reliance on Lists, Headings, and Structure
Ask a chatbot any question, and you'll get bullet points, multiple headings, and a conclusion section. This structure works in formal articles but feels forced in casual writing.
If someone asks ChatGPT whether a video game is worth buying, it responds with a short answer, then structured sections on "What It Actually Is," "Reception," and "When It's Worth Buying," capped with a conclusion. A friend or video reviewer wouldn't organize their thoughts this way.
6. A Lack of Unique Content
Large language models train on existing data and often search the internet for current information. They compile and rephrase what others have already said, rather than forming original opinions.
A blog post reviewing a new product becomes suspicious when it covers only generic information that matches what ChatGPT returns-the same information you'd find in multiple reviews. Personal anecdotes and unique observations suggest human authorship.
7. AI Detection Tools
Online tools like Copyleaks, ZeroGPT, and QuillBot analyze text and estimate the likelihood it was AI-generated. These services aren't reliable on their own-the same piece might get flagged by one tool and pass another.
If you use detection tools, run text through multiple services and cross-check results. Even then, don't treat the verdict as definitive.
Look for Patterns, Not Any One Thing
Every characteristic listed above appears regularly in legitimate writing by real people. Some writers naturally favor grammatical precision or structured formatting.
The key: if a piece exhibits most or all of these traits, it likely contains AI-generated content or was written entirely by a bot.
Don't alter your own writing to avoid these patterns. That approach will sound awkward and is unnecessary. Focus instead on being the best writer you can be.
Spotting AI text is a skill that requires practice, and AI models improve constantly. Mistaking AI content for human writing will happen-that's just the reality now. The telltale signs of AI-generated text are already changing.
To strengthen your ability to detect and use AI effectively, consider exploring ChatGPT Courses and Prompt Engineering Courses. Understanding how these systems work makes spotting their output easier.
Your membership also unlocks:
