Tips for Editing AI-Generated Writing in the B2B Industry
With the time savings offered by a large language model (LLM), many brands are now able to generate content that defines a distinct company identity. But unfortunately, many AI-written articles sound alike, and as more companies rely on AI to find their voice, unique perspectives may be drowned out by a cacophony of cookie-cutter LLM output.
This isn’t to say that you should avoid using an LLM. Rather, the key to standing out lies in how well you edit what an LLM provides you. Refining an AI-generated first draft into a compelling and unique published piece isn’t easy, but if you can marry the time savings of LLM use with the quality standards of good human writing, you can generate new fans while your competitors bore them.
By following the five tips below, you can quickly pinpoint where to focus your editing efforts to refine a piece so that it reflects well on your brand, builds search engine authority, and helps to build your site’s readership.
Treat AI Like a Friend, Not an Expert
Anything ostensibly factual in AI writing needs to be verified. AI is great for assembling “likely true” information, but it can’t be your final source. There’s no reason to assume it is inaccurate, but just like a piece of trivia a friend tells you, it needs to be verified before you can be sure. After all, you aren’t going to cite ChatGPT or its brethren as your final source, since it undercuts your site’s authority if users believe content is simply copy-and-pasted straight from an LLM.
This healthy skepticism goes beyond accuracy. You can’t necessarily trust the breadth and depth of research AI provides either. If you asked a friend to go do research for you and their end result seemed very superficial, you would probably figure your friend was in a hurry, rather than assuming only superficial sources existed. If the LLM’s coverage of a topic appears surface level, it probably is. Manually inserting more striking and relevant facts can help make an article better than those competitors generate with AI.
You’ll also have to check details like grammar, the validity of links, and perhaps most importantly, plagiarism, since LLMs are trained on preexisting content. Even if you don’t realize you’re submitting a plagiarized piece to a publication, robust plagiarism filters will likely detect it and your piece will be rejected. Even if you're publishing it on your own site, "your search engine rankings may be negatively impacted. So no matter where you want to publish a piece, plagiarism will hurt your search efforts and your corporate image, while wasting your time.
Ensure Your Article Is Properly Skimmable
Whether an article is written by AI or not, readers will continue to skim articles as they always have. This is a real problem for AI writing. LLMs tend to produce very similar introductions for similar prompts (as BuzzFeed found out the hard way). Now, more people are jumping from article to article, so the chance of seeing cookie-cutter openings is skyrocketing — as is the opportunity to earn readers’ attention by having a unique, compelling, and informative intro.
Similar attention should be paid to an article’s conclusion. Communications teams that need quick press releases or blogs may ask an LLM to hit a certain length. In response, LLMs can (and often do) pad out an article with a repetitive conclusion that adds nothing new while failing to properly summarize the piece for skimmers. Should an otherwise-stellar prompt result in such a conclusion, you may try reworking the prompt; alternatively, it may be beneficial to just write the ending yourself.
Finally, make sure that paragraphs begin and end in logical places. If a reader is sufficiently invested to graduate beyond skimming the intro and conclusion, they may still rush through the body. A poorly placed paragraph break may result in a jarring point of confusion for speed readers, making it obvious the content was written by AI.
Introduce Passion Through Lively Writing
As triumphant as it can feel to find the right prompt to generate a coherent article, coherence is not the same as passion. Editing should make an AI piece feel like there was real passion that inspired it — and ideally, at least a few readers will grow to be passionate about the topic themselves.
Passionate writers won’t use a monotonous approach to describe what they care about, and edited AI writing shouldn’t either. Employ sentences of various lengths and tones. Use a variety of synonyms rather than repeating the same words. Add real-life examples to drive your point home. And ensure that your work aligns with your company’s style guide.
Don’t Let AI Go Overboard with SEO Optimization
It’s hard to believe that hiding white keyword text on a white background was once a way to rank on Google. We’ve come a long way since then, but keyword stuffing is still a telltale sign of a marketer in training. And generative AI seems inclined to include many uses of any keyword it is instructed to optimize for. This is overkill; an article can rank with only two or three uses of a keyword, so an AI-generated article with far more uses of the keyword may actually hurt ranking while frustrating readers.
Be Honest If AI Is Not a Good Fit
ChatGPT and other LLMs promise to do for writing what the internet did for scholastic research — turning an hours-long process into a 30-second one. But editing AI writing is a time-consuming process of its own, and if the time loss is great enough, it may be preferable to simply write the article yourself.
To be efficient, you need to be able to tell early on whether AI can help save you time with any given article or corporate communication. Topics for which you may want to shelve LLMs include:
Technically complicated subjects for which small inaccuracies may frustrate a knowledgeable audience.
Sensitive subjects for which imprecise word choices could lead readers to assume the writer is not treating the subject with sufficient gravity.
Call-to-action pieces that suggest the reader should do something differently. Readers do not like being insulted or patronized, and achieving a positive tone while arguing for the value of one choice over another is delicate work.
While some writers may not be used to editing someone else’s work, editing AI will become standard practice for any company that needs to communicate at scale. Our awareness of the pitfalls of AI writing will grow naturally with experience, and we will have to acknowledge those flaws to become better editors. With the right combination of a great article topic and the right editing steps, you can quickly put out articles that stand out from the crowd and establish yourself as a leader in the era of AI writing.
