The safest place for help, and the rules for it
Editing prose you wrote is the one stage where assistance is nearly free of cost, because the argument is already fixed. It is also where a lot of students accidentally lose their voice, by accepting everything offered.
The governing rule: accept changes you can explain. If you cannot say why the new version is better, do not take it. That single rule keeps the prose yours and turns each edit into a small lesson in writing.
Ask for diagnosis, not replacement
Here is my paragraph. Do not rewrite it. Tell me: which sentence is unclear and why, where I have used two words for one idea, which sentence is longest and whether it needs to be, and where I have asserted something I have not supported.
Diagnosis, then you fix it. This is slower and it is the version where you get better at writing. A rewritten paragraph handed back teaches you nothing, because you cannot see the reasoning that produced it.
When you do want a rewrite for comparison, ask for it alongside yours with the changes named:
Rewrite this one sentence, then list every change you made and why.
Then decide, change by change. Usually two of the six are improvements and the rest are the model’s preferences, which are not yours.
What the changes usually are
Assistants edit towards a particular register: more formal, more hedged, more abstract, more evenly-weighted sentences. Some of that helps academic writing and some of it is exactly what makes prose lifeless.
Watch for these specifically, and reject them by default:
- Nominalisation. We investigated becoming an investigation was undertaken. Longer, vaguer, and it hides who did what.
- Hedge inflation. This suggests becoming this may potentially suggest that in some cases. Three hedges where one was honest.
- Elegant variation. Calling the same thing by three different names to avoid repetition. In academic writing, repeat the term; the reader needs to know it is the same thing.
- Sentence-length flattening. Everything becoming the same medium length. Rhythm comes from variation: a long sentence doing complicated work, then a short one landing it.
- Abstraction creep. Specific examples replaced by general statements. This is the most damaging one, because specificity is what earns marks.
The edits worth making yourself
The list that improves almost any student essay, in order of return:
- Cut the first sentence of every paragraph if it is throat-clearing. It is important to consider that... Start at the claim.
- **Find every it is and there are.** Most can be replaced with a real subject and verb. There are several factors which influence becomes Three factors influence.
- **Check every this has a noun after it.** This shows: this what? Ambiguous this is the most common source of unclear academic prose.
- One idea per sentence where the idea is difficult.
- Read it aloud. You will hear what you cannot see. This is the single most effective editing technique and it costs nothing.
Grammar and spelling tools
Spellcheck, LanguageTool (free, open source, and better than most paid options for non-English languages), and the grammar checkers built into word processors are ordinary writing tools and almost universally permitted. Grammarly’s free tier is similar; its paid rewriting features are generative and sit in a different category, and some institutions treat them differently. If your policy distinguishes, the line is usually between correcting an error you made and producing text you did not write.
For students writing in a second language, this category of tool is levelling and you should use it without guilt. Keep your drafts, for the reasons in Module 7.
The read-aloud test, one more time
If a sentence is hard to say, it is hard to read. If you run out of breath, it is too long. If you stumble on a phrase, so will your marker.
Every word processor now has a read-aloud function, which is free and catches things your eye skips because you know what it was supposed to say. Ten minutes of listening to your own essay before submitting will catch more than any amount of re-reading.
Every lesson here is open. Your progress is recorded once you pass the AI at Work exam.
Check your understanding
An assistant changes “We measured reaction times in 40 participants” to “Reaction time data were collected from a participant sample.” Why should this be rejected?
Pick the one you would defend. Your answer stays private.
Stuck on something?Ask Asha, your AI tutor
