So many of the social posts, blogs and newsletters I’m reading lately sound the same and, honestly, it’s exhausting. Same formula. Same closing line dressed up in slightly different words. If you feel it too, you’ll likely agree: AI is hurting communications more than it’s helping.
Staying human in the age of AI requires an editorial standard. Right now, many brands are losing those standards in favor of accelerating their content output.
Using AI tools well means being a better editor than a writer. Treat what comes back as a first draft that needs a senior editor, not a finished product that needs a light proofread. Most people who end up with slop stopped editing too soon.
That’s the mindset. Here’s how we apply it.
AI made everyone a writer. The job now is to be the editor.
Before AI, producing content at scale required skill, time, and judgment. Those constraints were a filter. They forced prioritization: what’s worth saying, who should say it, and why does it matter to this audience right now. AI removed the constraint. Anyone can generate a blog post, a newsletter, or a thought leadership article in minutes. The filter is gone.
What that means for marketing and PR teams is a role shift most haven’t fully made yet. The production problem is largely solved. The editorial problem is not. Writing is now the easy part.
Deciding what to keep, what to cut, what’s true to the brand’s voice and credible enough to publish under an executive’s name: that’s where the real work is. Teams that treat AI as a shortcut to publishing more will produce more slop. Teams that treat AI as a first-draft engine and themselves as senior editors will produce content worth reading.
That transition requires a different set of instincts. A good editor asks harder questions than a writer does. Is this claim substantiated? Does this section earn its place, or is it just filling space? Would the person whose name is on this actually say it this way? Does the argument build, or does it just accumulate? Those questions don’t come from the tool. They have to come from the person reviewing the output.
Start with something real
AI is most useful when it has something real to work with. The best outputs don’t start from a blank prompt. They start from a credible source: a client’s point of view, original research, a conversation with a subject matter expert who actually knows the territory.
AI doesn’t know what your brand believes. It doesn’t understand your market’s nuance, what your buyers are trying to solve, or which ideas your executives are willing to defend out loud. Strong thought leadership is grounded in what someone has seen, learned, or tested. A real thought leader can explain why a trend matters, where it’s headed, and what a buyer should do about it. AI can help shape that thinking into content. It shouldn’t be asked to generate the thinking itself.
One approach that works well: feed AI a transcript from a publicly available interview or podcast appearance and ask it to identify recurring phrases, the examples the expert keeps returning to, and the themes they care about most. Then let those patterns shape the content. A transcript-grounded draft sounds like it came from a person with a perspective. A prompt-only draft usually doesn’t.
Speak before you type
Voice notes produce better raw material than written prompts. When you dictate your thinking, you capture how you work through an idea: the emphasis, the detour, the point you return to because it genuinely matters. A typed prompt compresses all of that into something efficient and flat.
AI writing defaults to the mechanical. It finds structures that function well and repeats them. Dictated notes and interview transcripts carry something AI can’t manufacture: priority, rhythm, and the texture of someone who has thought about the problem. When AI has access to that kind of input, the output reads differently. When it doesn’t, you usually get something that sounds assembled rather than argued.
Whichever tools you use, someone still must ask the harder questions before the draft goes anywhere: Does this sound like us? Would our expert really say this? Does this add anything the audience hasn’t already read somewhere else?
Train it. Then keep training it.
A custom voice needs to be built from the right material and corrected over time. Start with content that already reflects who you are: approved blog posts, executive bylines, case studies, messaging documents. Pull in content from before AI became part of the workflow, especially material shaped entirely by human judgment. That’s the baseline. It shows the model what your brand sounded like before the outputs started converging.
From there, set parameters and keep refining them. How informal can the voice get? Should it read as authoritative or accessible? Does your buyer expect technical shorthand or does the brand always provide context? These are calibrations you should return to every time you produce a new piece of content.
The step most brands skip: feeding the edited version back. After you revise a draft, give the final version back to the tool and explain what changed. Where was the original too broad? What phrases were cut? What transitions were rewritten and why? That feedback loop is how a custom voice sharpens over time. The human edit is the standard the tool should be learning toward.
Trust is still built by people
AI tools will keep changing. New features, new platforms, new claims about what the shortcut looks like this time. Some of it will genuinely be useful. None of it changes what audiences are responding to when they decide whether your content is worth their time.
People do business with people they trust. That was true before AI and it’s more relevant now, when the volume of content is high and the cost of producing it is lower than ever. Buyers, journalists, and customers can feel when a piece was written by someone who knows the subject. They can also feel when it wasn’t. The content that builds relationships and strengthens reputation is the content where a real person’s judgment and experiences are visible in every paragraph: the claim that gets made, the example that gets chosen, the point that gets cut because it wasn’t important enough to keep.
Every output still needs to be questioned against what your brand or thought leader believes and can substantiate. A well-structured paragraph with nothing behind isn’t actually building any authority.
If you hold firm on your editorial standards and keep real people in the editor’s chair (regardless of how impressive the tools get), your content will continue to be read, trusted, and included in the conversations that matter to your business.
For more on how AI-generated content is affecting audience trust, this is worth reading.
