You're marking essays and something feels off. The structure is too clean, the vocabulary too consistent, and the conclusion reads like it was written by committee. You suspect a student used ChatGPT — but how do you check?

This guide walks through the process step by step, honestly and fairly.

Step 1: Trust your instincts, but verify

If you've been teaching for any length of time, you know what your students' writing sounds like. When an essay suddenly shifts in quality, vocabulary, or structure compared to a student's previous work, that's worth investigating.

But instinct isn't evidence. You need something more concrete.

Step 2: Look for the telltale signs

AI-generated text has consistent patterns that humans rarely produce:

Sentence uniformity. AI writes sentences of similar length and structure. Human writing varies wildly — short fragments, long rambling sentences, sudden changes in pace. AI text feels metronomic.

Hedging language. Phrases like "it is important to note that", "it's worth mentioning", and "one could argue" appear far more frequently in AI text. These are filler phrases that add nothing but sound academic.

Formulaic transitions. "Furthermore", "moreover", "additionally" — used mechanically at the start of paragraphs. Human writers use these too, but AI uses them as structural scaffolding in nearly every paragraph.

No personal voice. AI text is polished but generic. There are no opinions, no anecdotes, no moments where the writer's personality shows through. Everything is diplomatically balanced.

Vocabulary tells. Words like "delve", "multifaceted", "leverage", "tapestry", and "holistic" appear far more often in AI-generated text than in typical student writing.

Step 3: Run it through a detector

AI detection tools analyse text statistically and compare patterns against known AI outputs. No detector is perfect, but they provide useful signals.

When choosing a detector, look for one that does more than give you a percentage. A score alone isn't actionable — you need to know which passages were flagged and why.

Is It AI? combines AI analysis with pattern detection to highlight specific passages and explain what triggered the flag. That gives you something concrete to discuss with the student, not just a number.

Step 4: Compare against previous work

This is often more telling than any tool. Pull up the student's previous assignments and ask:

Does the vocabulary match?

Is the sentence complexity consistent?

Does the essay reflect topics discussed in class, or only publicly available information?

Are there specific examples from lectures, readings, or class discussions?

A student who normally writes in short, direct sentences but suddenly produces flowing academic prose is worth a second look.

Step 5: Ask follow-up questions

The most reliable check is the simplest: ask the student about their essay.

"Tell me about your research process for this."

"What was the hardest part to write?"

"Can you explain what you meant in paragraph three?"

A student who wrote the essay themselves can talk about their thinking. A student who didn't will struggle to explain their own arguments.

This isn't about catching someone out. It's about understanding whether the student engaged with the material.

Step 6: Have a fair conversation

If the evidence points toward AI use, approach the conversation with curiosity rather than accusation.

"I noticed some patterns in your essay that are common in AI-generated text. Can you walk me through how you wrote it?"

This gives the student a chance to explain. Maybe they used Grammarly aggressively. Maybe they're a non-native English speaker whose formal writing triggers false positives. Maybe they did use ChatGPT — and this is an opportunity to discuss why that's a problem and how to use AI tools responsibly.

What NOT to do

Don't rely on a single detector score as proof. AI detection is probabilistic, not definitive.

Don't accuse without evidence. A gut feeling and a percentage aren't enough for formal proceedings.

Don't assume malice. Many students don't fully understand academic integrity policies around AI.

Don't ignore it. Letting it go is unfair to students who did their own work.

The bottom line

Checking for AI use is now part of the job. The good news is that a combination of detection tools, manual review, and direct conversation is effective and fair.

Start with a detector that shows you why text was flagged — not just a score. Then use your professional judgment, knowledge of the student, and a fair conversation to reach a conclusion.

Try Is It AI? free — paste any text and see flagged passages with explanations in seconds.