Saturday, March 21, 2026

Using AI at Each Step of the Writing Process, Part 4: Revising

After drafting a rough copy of a business-critical document for a broad audience, a wise writer would run it through three rewriting steps: revising, editing, and proofreading. In short, when revising, we consider the content itself and the structure. Editing requires us to look at the fluency, clarity, and conciseness of the sentences; the word choice, punctuation, and spelling; and usage rules, such as capitalization, abbreviation, and numeration. During the proofreading step, we search for overlooked errors, such as typographical errors, spacing inconsistencies, and font irregularities. This post focuses on AI's effectiveness in revision.

We hear the word editor and proofreader much more than we hear the word reviser. When I use the word, many people will ask me, "Is there such a word?" It's just as well that we don't use the word much. An editor can edit someone else's manuscript, and the same goes for a proofreader. The fresh eyes of an editor and proofreader can detect opportunities to improve the quality of the language or fix mistakes. But writers should revise because only the writers know what they're trying to get across. They know the story they're trying to tell. Writers should know better than anyone else what information should be deleted, moved, or added to get their point across.

For this reason, I would not recommend AI to skilled writers for revising. They should know the content that goes into their message. For instance, if you were to write an accident investigation, you might think that a chronological approach would work. You would mention the cause of the accident and then the effect, followed by the response and mitigation plan. But skilled writers know the most important pointsto start with are the human and property damage. AI would know all of this, of course, so it would be useful to novice accident investigators. Moreover, most organizations require investigators to use a prescribed template for accidents, so using AI may be pointless.

But AI can give good revision advice if, after writing an internal proposal or evaluation report, you ask it questions like, "Should I delete anything? What should I add? Is the order of ideas all right?" It will give you helpful ideas to consider. As long as you remember that you, not AI, are in charge, you will find it useful during the revision step.