There’s endless discussion about how to prompt generative AI to produce great work for B2B content creators, but much less about the editing process after finding the optimal prompt. It’s high time that gets fixed, as there are always ways to edit large learning model (LLM ) output so it stands out in a field of AI-generated content. And as LLM work becomes more common, editing will separate compelling articles from generic ones to build readership and gain a competitive advantage.
Here are five tips for refining the output your generative AI assistant of choice provides you, to make that writing more effective for search engines and readers alike.
- Edit AI Like It’s an Unreliable Friend
You know who we’re talking about—the one acquaintance who claims to know a little bit about everything, but their insight falls flat after a talking point or two. That’s what it’s like to work with ChatGPT and its brethren. The work is probably accurate (kind of), but you can’t be sure, and even when the writing is factual, it’s pretty surface-level. That said, your friend probably doesn’t claim credit for famous novels and entire classes’ worth of student essays.
Because AI output has no accountability, you have to double-check everything down to the smallest detail—cited facts, links, grammar, etc. Because LLMs are trained on pre-existing content, there’s always a risk of plagiarism. Grammarly or another plagiarism checker is essential for avoiding Google penalties tied to copying others’ content.
- Nail the Intro and Conclusion
Even as AI upends content marketing, one truth remains: Readers will skim many of the articles they find online. And because they often bail after the first few paragraphs, your intro needs to grab and command attention. Unfortunately, LLMs will often produce similar openings for similar prompts, meaning unedited AI work will resemble other pieces you or competitors have written. Such a resemblance came back to bite Buzzfeed after a series of trite and nearly identical opening paragraphs made the site’s AI-written articles an easy target for critics.
An article’s conclusion will probably need a fluff-up, too, because LLMs frequently pad the ending of an article to hit a requested length. As students, we all knew the agony of being 100 words short on an essay and adding low-value content to reach the minimum length. For students, this merits a lesson about better pacing and developing more ideas. Writers relying on LLMs should either re-prompt for more unique content or manually add something interesting to the conclusion that ends with an appropriate call to action.
- Make the Writing Sound Like You
A piece of work can sound professional and knowledgeable without sounding like the writer is passionate about the topic. While editing an AI draft, make the piece feel engaging. Include a variety of sentence structures, replace repetitions of a word with more interesting synonyms, and add additional examples to tie the topic to real-world issues. An LLM can help generate an excellent foundation for the final product, but it’s your job to build something that can engage and excite readers. When in doubt, refer your company’s style guide. If one doesn’t exist, cross-check your content with existing articles. Get a feel for the tone of voice and replicate it.
- Recognize the Limits of AI SEO Optimization
It’s been a long time since Google tolerated aggressive keyword stuffing as a means of search engine ranking, but it’s a tempting strategy for any amateur content marketer, LLMs included. If you prompt an AI to write an SEO-optimized article, be on the lookout for excessive uses of the same term. Only two to three uses of a given keyword are needed to rank for it, provided other best practices are implemented.
Also recognize that SEO is often greater than the sum of its parts, and an article is only one factor that goes into that sum. A broad strategy spans many pages and considers historical performance, competitor strategy, and underutilized search terms. Even in a perfect world, an LLM couldn’t make a “perfectly optimized” article, since the content of an article is only a small part of the optimization process.
- Don’t Let ‘Saving Time’ Waste Your Time
If you asked a random person whether writing with ChatGPT or another LLM saves time, their answer would probably be, “Of course.” But the 30-second process of generating a draft from AI is only a part of the writing process. When you factor in the process of tweaking phrasing, checking facts, shoring up grammar, and other changes, you’re looking at a much longer writing process—perhaps even longer than writing from scratch.
If the article is especially complicated, demands nuance, or needs to achieve a precise tone, you’ll likely need to spend a lot of time refining an AI-created first draft. Be honest about whether you are actually saving time with an LLM or just shifting the workload from writing to editing and rewriting.
Acting on these five tips is a great start, but it’s only the first step toward producing great content and a great content strategy. AI is hitting the fast-forward button on content marketing evolution, so it’s essential to have an informed perspective so you can approach the latest trends with confidence. Reach out to Look Left Marketing if you want to find a solid brand voice, earn media coverage, build a prominent online presence, or establish yourself as an industry thought leader.