What do perplexity and burstiness mean in AI detection, in plain English?
I run the information literacy sessions in a secondary school library in Ireland, and next term I have a slot on "how AI checkers work" for our transition year pupils. Every explainer video they find says the same two words, perplexity and burstiness, and then moves on as if that settles it.
Cite
Global 100 Forum, "What do perplexity and burstiness mean in AI detection, in plain English?", https://global100.org/forum/q/what-do-perplexity-and-burstiness-mean-in-ai-detection-in-plain/, accessed 2026-11-15.I would like to explain both terms properly to 15 and 16 year olds without getting it wrong. My current draft says: perplexity is how surprising the words are, burstiness is how much the sentences vary, and AI text is low on both. Is that roughly right?
Two follow-ups I cannot answer myself:
- Do the big detectors still actually use these two measures, or is that an old description that keeps getting repeated?
- What should pupils take away from it? I do not want the lesson to turn into "write weirdly and you will pass", and I do not want them to think a low score proves anything either.
A short checklist I can put on a slide would be ideal.
Write something first.
Give people something to work with: at least 30 words on what happened and what you tried.
That is too long. Keep it under 6,000 characters.
Write the question as the title, 15 to 140 characters, no links.
Pick a category.
Add a name (2 to 40 characters, no links).
That email address does not look right.
That was quick. Read the thread, then try again.
The form expired. Reload the page and post again.
Something went wrong with the form. Reload and try again.
Please complete the check and post again.
Limit reached for now. Try again later.
This thread is closed to new replies.
This thread no longer accepts replies.
Something went wrong with the form. Reload and try again.
Post a reply or question first (name and email), then this browser can vote, edit and accept answers.
You cannot vote on your own post.
Only the author (within 30 days) or the forum team can do that.
That email already has an account. Use your sign-in link instead.
New accounts are paused for the moment. Try again later.
That was already posted.
1 answer
- Hana SatoMember ·
Your draft is close. Perplexity is measured against a particular language model: roughly, how predictable each next word is to that model. Low perplexity means the model would have guessed those words. Burstiness is about variation: whether that predictability jumps around across the text or stays flat.
The part I would add for pupils is the "to that model" bit. Perplexity is not a property of the text alone. Score the same paragraph with a different model and you get a different number.