AI Can Draft It But Can’t Finish It: Why Editing Still Matters
- Alex Bennett
- Aug 12
- 5 min read
Generative AI sets a trap for writers of every kind. Whether you’re using the technology to write an email or a full-length book, the right mindset is essential to avoid the snare.
The trap is easy to miss. It doesn’t seize your foot in a vice. It doesn’t zap you with electricity.
What it does is make your brain profoundly lazy. It hands you a warm blanket and a cup of tea laced with sedatives while whispering, “Go to sleep, my grammar is fully automated.”
Beware the passenger mindset
Writing with generative AI is like riding at the back of the bus. You get where you’re going, but you have no clue how you got there.
Unlike riding a bus, your destination with AI isn’t something fixed in the real world. It’s an amalgamation of words that an algorithm has strung together according to unknown (and unknowable) rules.
Some generative AI platforms are trained to cite their sources, which can help the user know where the words are coming from. Citations can be useful when you’re dealing with an unfamiliar topic, but at best, they are a backwards and myopic approach to research. Worse, tracing every source vaporizes your productivity. Before long, you'd have been better off manually researching and writing.
Most users won’t check the AI’s sources anyway. Ideally, they’re already experts in the topic, so they can spot inaccuracies and omissions in the generated text without asking deeper questions.
In my experience, that confidence makes experts especially vulnerable to the passenger mindset.
The Case of the Passenger CEO
Not long ago, I provided content strategy advice to the CEO of a small tech firm. For several months, he had used the base model of ChatGPT to generate marketing content. He reasoned that it was much cheaper than paying a marketing officer or hiring an outside agency.
After reviewing his content, it was clear to me that he was reading the first few lines of an article, skimming the rest, and publishing. He wasn’t driving the bus.
In his passenger mindset, he was missing a lot of important things about marketing content:
Core features of the company’s brand and values were missing, making the text feel generic.
The copy was rife with ChatGPT’s tendencies toward hyperbole and abuse of prepositional phrases.
Longer articles were repetitive and veered into irrelevant tangents.
Calls to action were vague and not tuned to the audience.
The text lacked a coherent SEO strategy.
A CEO has other things to worry about besides editing copy for content marketing. It’s understandable why this one—confronting an urgent business need under tight budget constraints—would opt for the machine.
The problem is that the goal of content marketing isn’t simply to publish for the sake of it. It’s to convey a message, deliver value, and build relationships. His generated articles weren’t accomplishing any of those things particularly well. They were watering down his brand.
Aside from delegating his content marketing to someone else, what the CEO needed was an editor’s mindset.
Drive genAI like an editor
Professional editors have an ideal skill set for working with generative AI. Along with knowing how to recognize and fix bad writing, they hold the key to escaping the passenger mindset: They have focus.
Many talented writers are mediocre editors, usually because they don’t enjoy it. Few academic programs teach editorial skills, and most professions don’t rely on them. Lawyers are one exception: even if they hate editing, they are obligated to do it.
Editors are the odd ones who love to focus on a text.
If you’re working with AI to generate communications of any kind, it pays to learn how to muster an editor’s focus.
(I recommend following my blog here at Amplitude Strategic if you’d like to learn more about the nuances of editing AI-generated text.)
A good editor pays attention to numerous details that distinguish a mediocre text from one worth reading. Here are some examples:
Start-to-finish cohesion. An article needs to deliver value in every paragraph, or readers won’t come back.
Fitness for purpose. Content typically has a main objective, expressed its call to action, but it also has other goals, such as expanding awareness of a company’s capabilities or touting its values. AI tends to make arbitrary choices to achieve technical goals, like adding SEO terms in places that don't make sense. An editor restores balance.
The logic of common sense. AI makes downright weird connections sometimes—things a human would never write or say. Correcting these issues is vital, but spotting them isn’t always easy. (Check out this article to learn more about spotting the problems with ChatGPT’s abuse of correlative conjunctions.)
Audience alignment. The right words and phrasing are paramount for holding an audience’s attention and conveying authenticity. Your content needs to reflect the culture of your industry.
Building the bigger picture. An individual article is unlikely to launch an empire. Instead, it contributes to a growing library of related content that together forms a durable asset.
Many of these examples call for applying abstract concepts to a text. Generative AI isn’t necessarily bad at this stuff; at a minimum, it’s better than a human who isn’t a competent writer. But your content standards should be higher than mere competence.
Editing AI is a skill you can learn
Like any skill, editing can be learned. Anyone who intends to rely on generative AI for important communication should cultivate an editor’s mindset. These are some ideas worth thinking about:
Slow down. Patience is the most powerful counter to the laziness trap. Just because the AI moves at inhuman speeds doesn’t mean we should, too. Resist the temptation to publish the first thing the AI generates. If you're writing an entire article in an hour, you're still working quickly.
Read drafts with purpose. Practice reading the entire generated draft as though it will be published on the front page of the New York Times with your company’s name in big, bold print.
Treat the AI like an intern. Generative AI is no better than a talented but inexperienced college kid. Don’t trust it to make good choices. Remember: You are the expert.
Set your standards. Every company should have a style sheet. It doesn’t need to be complicated or incredibly detailed, but it does need to set the rules by which your content is judged before publication.
Read a book on editing. If you learn well from books, there are numerous great guides to the editorial craft. Along with being a practical reference, Amy Einsohn’s The Copyeditor’s Handbook is an accessible and concise introduction. Scott Norton’s Developmental Editing is a useful overview of how to construct a quality draft.
Explore professional resources for editors. The Editorial Freelancers Association (or EFA for short) is an indispensable source of information and education for anyone serious about becoming a better editor.
Take a class. Classes in copyediting are the ideal way to build skills. Along with organizations like the EFA, many colleges offer online programs.
To get better results with generative AI, start with me
You can also hire a consultant like me. Along with providing writing and editing services, I offer group or one-on-one sessions focused on the skills that are most relevant to developing content using AI.
Generative AI is not a magical replacement for human creativity. It can, however, speed up your content development when used carefully.
If you haven’t already, be sure to subscribe to this blog to be notified when I post new tips. You can also reach me through my contact page or by emailing me at alex@amplitudestrategic. Let's put your content on a solid foundation for the long term.
Comments