The human element: avoiding over-reliance on AI in content creation

In the rapidly evolving world of content marketing…

Nah, just kidding. No AI was used in the making of this intro.

But, whether we want to or not, we content marketers have to wrangle with the implications of AI.

• Should we use it for outlining?
• Can we use it to write a first draft?
• If not, why not?
• If yes, how much AI is too much?
• What will using it do to our creativity and the quality of our writing?
• How can we take advantage of AI as a force multiplier, without turning it into a crutch?
• And, most importantly, are we still allowed to use the em dash?

Let’s dive in. 🚀🚀

Does it matter if your content sounds human?

Brooklin Nash, on X.

Much of the debate on using AI to generate B2B content is based on the assumption that “sounding human” is important for content.

Content marketers wax lyrical about content that’s engaging for a human reader. ‘You can’t just do it with AI,’ they say, ‘because then it will be boring and regurgitative, and that will be bad for your brand.’

There’s just one problem, though. Most content, especially B2B content, and even more especially search-focused B2B content, is not engaging for human readers. And it never has been.

It is dry, pompous, and startlingly unoriginal. As Ryan Law put it back in 2019, way before ChatGPT came along: “Explore the search results for virtually any lucrative keyword, and you’ll find a bunch of articles with lookalike titles, headers, and examples.”

The LLMs are a mirror. They show us what we really look like. Frankly, some content is so poorly written and hard to read that “sounding like AI” would be a step up.

In this context, does it really matter if your content sounds like AI?

Well, actually, yes. And this is the good news about generative AI.

B2B content marketers got away with low-quality, generic, SEO pap because there wasn’t that much competition. As Ronnie Higgins, content marketing leader and founder of Neutral Ground Labs and Marketing Under The Influence, wrote in a recent newsletter:

“B2B marketing was like shooting fish in a barrel in the early days. Because almost everything else on the web was a few years old, I could publish a blog post and see it rank in the top of SERPs the next morning.”

But now, everyone can afford to play that game. The playing field has been leveled. And that’s the good bit. Because, in a sea of sameness, standout content becomes the only sensible long-term approach. As Tim Metz, the Director of Marketing and Innovation at Animalz, puts it, “If everybody can spin out a thousand SEO articles every month, then just do the math. People only have so much attention.”

AI-forward content may indeed work right now, and may keep working for a while. But logic dictates that in the coming years, brands will only be able to stand out with content that feels obviously human. And standing out is going to become increasingly important as more and more bland robotic content gets pumped out.

And therein lies the opportunity for content marketers.

What does human content look like?

At this point, it may be worth defining what makes content feel human-written. Lists of “terms that make you sound like AI” abound on LinkedIn. They’re even coming for the em dash.

However, as Kiran Shahid points out in an article for Zapier, AI detectors are wildly inaccurate. And content that sounds human has features that have nothing to do with hyphens or the word “utilize.” Shahid suggests that human content differs from AI content by:

• Using a “what-why-how” framework. AI tends to skip out the part where humans naturally explain why something is important and why the reader should care about it.
• Introducing bold, subjective opinions based on expertise. AIs tend to hedge their bets by using neutral, non-specific language.
• Using punchy, direct language. AI tends to lean into jargon and meaningless filler phrases.

Human content is not the same thing as high-quality content, of course. Human content can fall anywhere on the quality spectrum.

The point is the potential of human content. It can be mediocre — but it can also be spectacular.

AI content can only ever copy what has already been said. Human content has the potential to be creative. Humans are capable of original thought. Humans can make unexpected leaps of imagination. We can ask non-obvious questions. We can think of unique examples. We can draw on our own experience, and our empathy for one another, to create content that is meaningful, astute, and genuinely useful.

AI can be pretty good. And that’s all it can ever be.

The risks of over-relying on AI for content

The biggest risk of using AI to write your content is that it will mean your content doesn’t work — that it doesn’t drive conversions and doesn’t improve brand awareness among your target market.

For instance, Edelman’s 2024 B2B Thought Leadership Impact Report found that 90% of B2B software buyers said that they would be “more receptive to sales or marketing outreach from a company that consistently produces high-quality thought leadership.”

When asked what “high-quality” meant, 66% mentioned content that “has a unique format or style that looks different to other thought leadership you see.” Or, to put it another way, content that would be almost impossible to create with AI.

Worse, “uncanny valley” AI content might actively discourage your market. Kane Jamison, the founder of Content Harmony, comments:

“People still have a negative reaction to content that feels human-like but isn't quite there. As soon as the reader picks up on that while reading your content, many readers feel disrespected and walk away dismissive of the site or author.

That might be OK for certain types of utility content, but for anything requiring expertise or personality (most branded content), it's a net negative customer experience for the brand.”

Too many organizations have been treating all long-form content as what Jamison calls utility content. But blog posts, deep dive guides, and data-driven reports shouldn’t be utility content.

Blog posts aren’t the same as a user manual or a product description. They (can) have a transactional function, for sure. But they’re also a great opportunity to establish your voice, showcase your expertise, and develop a relationship with your target market.

Writing them robotically (with or without AI) is wasting that opportunity, and minimizing the ROI of your long-form content effort. It’s putting your highly skilled content team on the same level as anyone with internet access.

Of course, wasting a great marketing opportunity by sounding like a robot isn’t the only risk of using AI to write your content. For starters, it can cost you your job (or your clients). Many content teams actively screen for obviously AI content. But there are other, less obvious costs:

1. Your creative muscles can atrophy

Starting at a blank page is good for you, says Metz. Reach too quickly for AI to “just get you started,” and you lose your ability to think for yourself, he wrote on the Animalz blog:

“I’ve had several periods where I tried to let AI do my writing. To be fair, the writing of recent models like Anthropic’s Claude is impressive and will only improve. But my brain stopped working. My ideas became a trickle, I fell for hallucinations, and I started relying on AI for even the easiest of sentences.”

The Animalz blog.

SEO strategist Olivia Barrow says that ChatGPT can help “cover the jobs I'm weaker at and let me focus on my strengths,” but that help comes at a cost.

“By not practicing the things I'm weaker at (the headlines, for example), I think I'm getting weaker. I would feel rusty at some of these things if you took ChatGPT away from me, the same way I am rusty at calculating 13 times 8 in my head because I have been using a calculator for so long.”

2. You lose your eye for quality

Another risk of over-relying on AI is that you stop being able to judge content quality. If you read enough AI fluff, it can start to look OK to you.

Metz recalls: “There was a period where I started to use more and more AI, and I started to abandon my previous process. And I’d read an AI article and say to myself, “Oh, this looks pretty good.” But then, if I took the time to look at it with a fresh mind the next day, I’d be like ‘Oh, no, this is actually pretty bad.’ From a distance, it looks good. But then you start to read it and you realize it doesn't mean anything.

I really had to reset myself. I even had to go back to my old process of doing the hard work, doing the research, organizing everything. No more just saying, ‘ChatGPT, summarize this, summarize that.’”

Using AI to get you off the blank page may speed up your writing process, but it anchors you to a fairly mediocre level of content quality, says Metz: “It kind of blinds you to the other possibilities and directions that could exist, but you don't see them anymore because what is in front of you seems pretty good.”

This is a risk, even if you just use AI to outline, says Jamison. Content quality suffers “when the creator starts letting AI make meaningful decisions,” and that includes “letting the AI develop the entire outline without thinking through each section and whether it belongs and fits the brand's voice and stance.”

3. You lose your opportunity to think

To quote Pulitzer Prize-winning author David McCullough, “Writing is thinking. To write well is to think clearly. That’s why it’s so hard.”

If you outsource writing to AI, you’re losing an opportunity to form your position and build your expertise on the topic at hand, says Erica Schneider, former Head of Content at Grizzle and founder of Cut the Fluff.

“You get clarity through the process of figuring out the messy sh*t in between your brain and your fingertips. And, if you outsource the clarity-seeking part of thinking, you're never going to go a level deeper in your own expertise. It's a terrible idea to let AI do the thinking for you.”

AI guardrails for content marketers

This isn’t an argument for rejecting AI outright. AI can be an excellent resource for content marketers. As Barrow puts it, “By giving me more time for the good parts of writing, generative AI probably helps me be more creative.”

It’s more a case for putting rules in place to help you get the most out of AI, while still being able to operate at full creative strength when needed.

Here are our experts suggestions for using AI without breaking your brain:

1. Don’t start with AI

For Metz, the first rule of using AI in content is that you don’t start with the AI. Start with your own ideas, insights, research, and thought process. Then use AI to help you refine and improve on what you’ve started.

“I think both the human and the AI do worse if you start off with the AI, because the AI will also do better if you give it better input. If you start with the AI, the scope of what the AI will come up with is limited.

Whereas if you start with human input, you do some thinking, so you have more thoughts about the whole concept. And then the AI will also do better because you send it off with something unique, instead of letting the AI take the first step, which will always be based on what it already has.”

Vahbiz Cooper, the Digital Marketing Manager at Demandbase, agrees: “If you rely on AI to generate everything without adding your own voice or perspective, it might stifle creativity in the long run. Like a trampoline, it’s great for a boost, but you still have to jump.”

2. Use it for content strategy

By focusing on how to write with AI, we may be missing out on far more useful applications earlier in the content workflow.

For instance, Higgins has been using AI since 2019. These days, he primarily uses it for research and development.

“At OpenPhone, our ICP was very broad: tech-savvy small businesses who relied on phone calls and text messages to build relationships with customers/clients. This meant the specific pains/gains related to the product were difficult to triangulate on my own and would have required an expansive (aka expensive) research project.

But when I gave AI examples of the ICP, details about OpenPhone, and asked it to help, we were able to pinpoint category entry points they all had in common.”

3. Use it as a writing partner

AI isn’t a great writer, but it can be an outstanding writing partner. Barrow says she uses AI “like an advanced thesaurus, where I'll describe a word I'm thinking of, or put in a bunch of slightly incoherent phrases and ask it to suggest a concise way to communicate my message.”

Confession time: that intro? Where did I say I didn’t use ChatGPT? I just realized I actually lied to you. I couldn’t remember the correct wording of the expression “force multiplier.” Was it a “force magnifier?” (What can I say? It’s January.) ChatGPT, on the other hand, had no difficulty whatsoever.

4. Never, ever skip the editing

Schneider is clear: “Don't have AI write for you. But, if you are going to have AI do all of the writing for you, you have to then deeply edit it.

But you should deeply edit yourself anyway. Editing is never okay to skip. Period. End of story.”

How should you edit it? Molly Weybright, a Senior Digital Content Strategist at Service Direct, says,

“My number one piece of advice for making AI-generated content sound human is to read the copy out loud. I've found that AI content generators can be wordy and use tons of adjectives. In short, they often read like a student trying to hit a word count. But, if you read the piece out loud, I think most writers will find obvious places to remove and rewrite copy to make it sound more natural.”

5. Use it for utility content

And, of course, AI is perfect for basic content where there’s no real need for creativity, original thought, or the human touch. For instance, Higgins suggests that technical documentation is an ideal use case for AI:

“Instead of writing a technical manual myself, I can give an AI all my notes, an audio recording of me rambling on how to use it, and pictures of the product for it to create something faster. As with anything AI generates, I'd still need to review it and ensure it didn't make sh*t up or confuse any details from sources.

That example of using AI works because the technical manual only needs to be technical — meaning it doesn't require a lot of empathy. But as far as creating content that fulfills needs beyond informational needs (emotional, social, spatial, and temporal), that'll require more human ingenuity and creativity.”

For more pointers on how to use AI — and how not to use it — check out our guide: Beyond AI: The Real Skills Content Marketers Need to Stay Relevant

Over-reliance on AI is not the problem — it’s a symptom

Finally, it’s worth exploring why so many content marketers have fallen head-first into over-reliance on AI.

Barrow really hit the nail on the head when she commented: “I don't think generative AI kills my creativity — being forced to produce too much content too fast is what kills my creativity.”

The average content team consists of 1-5 people, regardless of company size, according to Tommy Walker’s State of Discontent Report. That handful of human writers are expected to manage 5+ different content formats. They’re struggling to keep up with the demand for more and more content.

Meanwhile, they are often being judged on content volume or search rankings — both of which make generative AI the obvious tool for the job.

If we want content creators to use AI cautiously and use their own imaginations to develop original and creative content, then the solution is not to try and “catch them out” with AI detectors. The solution is to give them time to think. And then measure their content on its ability to delight, engage, and convert an audience.