3 Things You Must Do Before Sending Me AI-Generated Content

by | Content Writing and Copywriting

Dearest marketing manager,

I’m sorry, I have bad news. I know you have the best intentions, but it does not actually make my job easier when you send me AI-generated outlines, research and drafts of content to work with.

Two reasons for this:

  1. All generative AI tools hallucinate — yes, even when they cite their sources. An ethical writer locates the original source and verifies every fact.
  2. The AI doesn’t know your customer like you do (like I hope you do), and it did not create it with them in mind — yes, even if you specified your buyer persona in the prompt. A good writer will put themselves in your customer’s shoes and write to them, not at them. It’s a subtle difference, but it makes all the difference in how your content lands.

Now, I don’t want to discourage you from using AI to make your job easier. It has a lot of uses, and personally, I love collaborating with ChatGPT to bounce ideas around and make my own work more creative. (I delivered an entire live workshop on this in spring 2024!)

However, there are some “rules of the robo-road” that I give to my clients when they want to send me AI-generated documents, and I’m going to share them here with you in this article.

I share these rules with my wonderful clients to make sure their time isn’t wasted, and we are able to continue to produce the most meaningful, effective copy and content without costing them a fortune in added research and editing fees.

Here are three things I ask my clients to do before sending me or my team any AI-generated content:

1. Use at least three follow-up prompts.

The biggest mistake I see marketers making is that they ask an AI tool to produce something (an outline, explanation, summary, draft, whatever), and they call the output good.

Folks, even God took seven days to create the world.

Take another pass.

In fact, take at least two more passes.

After your AI tool has given you output, read through it and give the AI further directions in follow-up prompts. These directions can be things like:

  • Rewrite that for [specific and narrow persona].
  • Please make that more conversational and use less jargon.
  • What are two other ways you could structure that?
  • Please check this for accuracy.
  • Make that simpler / easier to understand / more compelling.
  • How would you explain this to a 13-year-old? Then how would you explain it to a 5-year-old?

Why do this? First, it helps you hone in on exactly what you want and need in the content. Second, it gives you other output to compare with the first “draft” output. And third, it helps you spot gaps and opportunities before you pay a writer to do that.

2. Check all the cited sources, or factor in the cost of having your writer do this.

This shocks most people, but when an AI tool gives you a link to the source of the information, you still have to check it.

I’ve lost count of the number of times I followed a source link, only to discover that information was nowhere on the page — or the page didn’t exist at all.

At Horizon Peak Consulting, we always check our sources. In fact, I give my writers very strict rules about this, and we follow them diligently. My writers are to 1) check the source of any facts or data they want to include in a piece of writing, 2) make sure they are citing the original source (too many articles cite other articles that cite other articles), and 3) do this in 10 minutes or less. If it takes an HPC writer more than 10 minutes to locate the original source of the data, they are not allowed to use that data in the writing.

This isn’t overkill, it’s ethics. We refuse to contribute to the rampant misinformation out there, and we won’t allow our clients to contribute to it either. We care too much about your relationship with your customers to let you break their trust like that.

3. Be okay with your writer changing things.

If you are working with an experienced writer — especially one that has a proven process, like we do here at HPC — they are going to know better than you. And I’m not sorry to say that, because that’s why you hired them. If you had the time to get as good as they are, you wouldn’t need to hire them in the first place.

So trust your writer. It’s their job to create a connection between you (your company) and your customers. AI has no such imperative.

If you send them an outline, and they want to rearrange it, let them. If they have a different idea for a topic angle that would work better, listen to it. You can (and should) ask why, and a confident writer should be able to explain their reasoning to you.

A writer who knows their craft, who has experience working within your industry and/or with your target customers, will have a feel for what will work. A feel that AI can’t replicate.

It’s a bit of human magic … just go with it.

Generative AI is great — but it has limits

I have been teaching on AI topics for almost two years, now, and I continue to be amazed at how little audiences really understand about the limitations of this technology. I’ve been horrified watching people copy AI output and publish it as their own, misinformation and all. I’ve watched companies’ traffic numbers and revenue tank as audiences lose trust in them by the second.

And … I’ve watched authors’ eyes light up as they give ChatGPT the role of a character in their book, and work out the kinks in the dialogue for a scene they’re writing. I’ve watched marketers communicate more clearly with team members because they now have tools to show what they have in their minds.

I’ve even watched beautifully human narrative elements find their way back into marketing, PR and sales communication — slowly but surely.

We are still very much in the hype cycle of generative AI. More is going to change. But one thing won’t: the deep desire humans have to trust and connect with one another. If we’re going to continue to use these AI tools in marketing, dear clients, I insist we do it with trust and connection in mind.