Skip to content
Sound More Human

For educators

What this tool does, what it cannot tell you, and a fairer way to have the conversation when you are unsure who wrote something.

Writing as a student? Jump to the note for students.

What it does

Someone pastes text into Sound More Human. An AI model rewrites it for varied sentence structure and a more natural rhythm, and the rewrite appears beside the original with every changed word highlighted. Students use it, and so do job applicants and anyone with a stiff draft. The text it starts from may be a person’s own draft, or it may be machine-written; the tool does not know which.

What it cannot show you

  • Who wrote a draft, or whether AI was used. Neither the tool nor its output can tell you. There is no report to request and no way for us to confirm that a given piece of text went through it; what a person submits stays private to their account.
  • Anything about AI detectors. We make no claim about detector scores, and a detector score is not evidence of authorship. The research on how often detectors are wrong, and about whom, is summarised in Are AI detectors accurate? What the research says.
  • That a rewrite kept the meaning or the facts. Nothing checks it, and a rewrite can change what a sentence claims. The highlights show which words changed, and that is all they show.

A fair conversation with a student

If a piece of work does not sound like the student, the machine-written explanation is only one of several. A conversation gets you further than a verdict, and it protects you if you turn out to be wrong.

  • Ask them to walk you through it. The argument, why they structured it that way, where each source came from and what it says. Someone who wrote the piece can usually do this; someone who did not usually cannot.
  • Ask to see the process. Notes, an outline, earlier drafts, and the version history that Google Docs, Word and most editors keep. Ask for it kindly and early.
  • Compare with their earlier work in the same course, knowing that a student’s writing changes across topics and across a term.
  • Take extra care with second-language writers. Plain, even prose is what detectors tend to flag, and the research linked above found them wrongly flagging writers working in a second language far more than others.
  • Decide on evidence you could explain to the student, and to a committee if it comes to that. A tool’s score is not that.

Check the course policy, and say it out loud

Whether a tool like this is allowed is decided by your course and your institution, not by us. If the syllabus says nothing about AI tools, students are guessing, and a sentence on the first day about what is allowed and how to disclose it prevents most of these conversations before they start.

For students

  • Keep your drafts. Turn on version history in the editor you write in, and keep your notes and outline. If anyone ever asks how you got to a piece of work, that is your answer.
  • Check your course’s AI policy before you use any tool on coursework. It is usually in the syllabus; if it is not, ask.
  • Disclose it if your course requires that. If the policy says to say when you used a tool, say so, in the way it asks.
  • Read the rewrite before you use it. It can change what you meant or add something you did not say, and the work you hand in is yours to stand behind.

More on what the tool does and does not do is on the about page. Questions from an institution can go to the contact address.