How AI writing tools actually work
Straight answers to the questions students actually ask before they trust a writing tool with an assignment.
Why does ChatGPT use so many em dashes?
Nobody outside the AI labs knows for sure, but the leading theories point to old books and how models learn to write. What is known, and what to do about it.
What is :contentReference[oaicite] in ChatGPT text?
Those strange strings are pieces of ChatGPT's citation system that survived the copy. They mean nothing outside ChatGPT. What each one is, and how to clear them without losing your sources.
Can you remove an AI watermark from text?
Hidden characters come out in seconds. The watermarks Claude, ChatGPT and Gemini now use do not, because they are the words themselves. What actually changes them, and what to watch for in tools that promise more.
Can AI edit a Google Doc directly?
Yes, but an in-document edit, a connected Google Drive action, and a new exported document are different workflows. Here is how to choose.
Is Write It worth it?
We wrote this one, so read it accordingly. Here is the honest version, including the cases where you should not pay us anything.
What is Write It?
A complete, plain description of what Write It does, where it runs, what it costs, and what it does not do — no marketing adjectives.
What is an auto typer?
A plain explanation of what an auto typer is, how the good ones simulate real typing, and the honest limits of what that accomplishes.
What is an AI humanizer?
The specific things that make AI writing sound like AI writing — and what a humanizer does about each of them.
Google Docs version history: who can see it?
Who can open version history, what earlier edits show, and how access differs for editors, commenters, viewers, and teachers.
Does Turnitin detect ChatGPT?
Turnitin reports an estimated percentage of a document that reads as machine-written. What that number is, and is not, deserves a straight answer.
How AI detectors actually work
Every AI detector measures roughly the same two things. Understanding them explains both why detection works at all and why it misfires.
Claude and ChatGPT now watermark their text: what changed and what it means
Anthropic and OpenAI now mark the text their models write. Neither mark is a hidden character: both live in word choice. Here is how they work and who can detect them.
Is using AI for schoolwork cheating?
It depends on your school, your class, and the assignment — and "it depends" is not a dodge. Here is how to work out where the line is for you.
Write It pricing and tokens, explained
Free, Plus, and Premium, what a token is, what each feature costs you, and how to decide without overpaying.