Staring over the edge of my laptop at a stack of board games, I beg my brain to deliver an engaging lead for this blog.
It refuses. Not because it doesn’t want to, but because it can’t. Not as easily as it used to. Not before November 2022, anyway.
That’s when OpenAI publicly debuted ChatGPT, and every writer on Earth panicked, then embraced, and then quickly overused the company’s large language model (LLM) for both tiny edits and big ideas.
By now we all know LLMs, like ChatGPT and Microsoft’s Copilot, are advanced forms of artificial intelligence trained on vast amounts of text data to understand and generate human language in a highly sophisticated manner.
It didn’t take long for AI and content marketing to become joined at the hip. AI’s ability to help marketers scale production and repurpose existing content more efficiently quickly became indispensable. In fact, 90% of content marketers plan to use AI to support their content marketing efforts.
That said, most professional writers I know do not leverage LLMs to create from scratch. We collaborate, of course, asking AI to suggest outlines, rework bumpy sentences, craft emails, atomize articles into social media captions, analyze our work, and sometimes buoy our egos.
Yet, even with that collaboration, it seemed the more I relied on AI to tie up my loose ends and button down my messaging, the more I noticed something unsettling.
Slowly, it became a little harder to connect the dots creatively from one thought to the next. Ideas that once flowed effortlessly required more mental work. Writing felt muddy, like I slogging through some kind of neural sludge.
Was I the only one experiencing this?
Maybe if I reached out to Cathy McPhillips, Chief Marketing Officer at SmarterX, she would know.
But before I could type up an email and hit the send button, researchers at the Massachusetts Institute of Technology (MIT) published a study titled Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task.
To my surprise, people were experiencing the same neural sludge. Researchers called it cognitive debt.
Chatting with Cathy would have to wait.
Cognitive debt is not interest-free
Cognitive debt, if you were wondering, describes the reduction in mental performance that occurs when external systems (such as AI or search engines) perform tasks in which our brains would otherwise engage. Over time, this can impair memory, problem-solving abilities and creativity, as the neural pathways that support these skills are underutilized.
To better understand this phenomenon, MIT researchers designed a study in which participants were divided into three groups and asked to write an essay using different tools:
- One group used an AI language model
- Another used traditional search tools
- The third relied solely on their knowledge and brainpower
During the first three sessions, each group stuck with its assigned method. In the last session, the groups swapped tools.
The findings were striking. Participants who relied on AI consistently underperformed compared to those who used search engines or no tools at all. Their brain activity was weaker, their essays felt less “their own,” and they even struggled to remember what they had written.
While AI offered quick results in the short term, it appeared to reduce mental engagement and creativity over time.
MIT researchers are quick to point out that their work isn’t complete and will need peer review before conclusions are drawn. They’re also adamant that using AI does not reduce intellect.
But boy, does it feel that way. So what do I do?
Time to call in the experts.
How did I run up all this cognitive debt?
The lure of AI is strong. It can facilitate idea generation and research design, improve content and structuring, enhance data management and analysis, and assist with editing, review and publishing, according to one study.
But with it comes the risk of building up cognitive debt, which happens slowly and often goes unnoticed, says McPhillips.
“Faster outputs and creative efficiencies are enticing at first, but users soon sacrifice deep thinking and all that comes with it: in-depth analysis, critical evaluation and reflection,” she says.
“This is the risk that worries me most, because it’s the one we’ll see last,” adds Robert Rose, founder and chief troublemaker at Seventh Bear. Rose explores this more in his new book, Valuable Friction.
Cognitive debt isn’t about bad content. It’s about eroded judgment, he explains. It’s when we stop thinking critically, not because we can’t think critically, but because the system makes it easy not to.
“This isn’t a speed problem. It’s an epistemological one,” Rose says. “Epistemology, at its core, is the philosophy of how we know what we know — not just whether something is accurate, but why we believe it’s justified. That’s what’s at stake when AI hands us convincing content without context.”
Can you pay off your cognitive debt?
So, how do we best purge the neural sludge? Go cold turkey? Wean ourselves from the LLM?
How do we pay off our cognitive debt?
We could start by avoiding the debt altogether, says Rose, though it requires full-on cultural reframing. Rather than using AI to go faster, creators should leverage its abilities to go deeper.
“When AI offers an idea, a strategy, a suggestion, our role isn’t to accept it. It’s to pressure-test it, add context, and decide what would make it worth believing,” he says. “That’s not slowing down — that’s the real work of thinking.”
Regular content, creative and operations audits might help, too, says McPhillips.
“Are we starting from scratch too often with AI? Are we questioning its suggestions, or assuming it’s good as is?” she says. “Are our writers feeling less confident in their instincts?”
By bringing teams together for debriefs not just on what performed well, but how it was made, we can nurture a culture of curiosity and challenge, even when AI is doing some of the heavy lifting.
The flip side might be accepting our fate when trying to pay off that debt, says Joe Pulizzi, the godfather of content marketing and the founder of Content Entrepreneur Expo and Content Marketing Institute.
“Because I don’t think you can,” he says. “The more we use AI, the more we’ll lose the writing brain. The good news? Hopefully, we will use that cognitive talent for something different we never thought of.”
Is there a better way to invest in AI creatively?
In the study, writers who relied exclusively on AI tended to engage in “cognitive off-loading,” producing work that was noticeably more uniform — and often homogenous and dull.
But let’s be honest. Knowing that creators won’t abandon AI anytime soon, especially in fast-paced agencies and corporate content marketing teams, where the demand is always to be fast and deliver more with fewer resources.
With that in mind, is safeguarding originality and preserving brand voice even possible? If so, how?
First, understand that AI is very good at pattern recognition, especially complex patterns, but terrible at relating human expertise and experience, McPhillips says. That requires context recognition, nuance interpretation, and using knowledge with flexibility.
“It doesn’t understand who you’re writing for, and what you’re writing about,” she says. “When that happens, your content sounds like everyone else’s and it gets treated like everyone else’s.”
McPhillips’ advice:
- Infuse AI prompts with your point of view
- Start with a story only your team could tell
- Use brand-specific language
- Pull in audience-specific insights
- Challenge the model’s output if it feels too safe, off-brand, or “too AI.”
“When individuals or teams default to AI too soon, they may unintentionally stunt their thinking and default to generic patterns, or worse, remove critical subject matter expertise,” explains McPhillips. “Instead, I encourage marketers to bring their human ideas to the table first, study up on the topic, then use AI as a partner to sharpen structure, expand on ideas, or consider variations. AI should elevate human creativity, not remove it.”
Pulizzi’s advice is much more concise. “Don’t use it for writing,” he says. You could consider AI as a tool you can use to critique your first draft but do everything you can to preserve the act of creation.
“Much of the writing process occurs in the journey. The thinking of topics. The consternation over using the proper word. When your brain unveils a connection you didn’t think was there,” says Pulizzi. When a writer leverages AI to offload many of these “chores,” sometimes the writing falls flat.
“But even worse, the next time we begin the writing process, we’ll need that AI crutch again,” Pulizzi says. “After all this, what does writing become?”
Conclusion? Proceed with curiosity
Given the real sensation of atrophied skills that led me to this post in the first place, the answer to Pulizzi’s question is profound, whatever that answer may be.
Human creativity allows us to connect ideas, solve problems, and produce unique and useful things, from artistic expression to emergency medicine and everything in between. It thrives on the struggle. On wrestling with words, teasing out meaning. By handing that over to AI, we lose the experience creativity gives back.
Surprisingly, that doesn’t mean there isn’t a place for AI in the creative process.
Though research is ongoing, preliminary results from the MIT study suggest that there could be ways to balance AI efficiency with human originality.
To that end, Rose suggests treating every AI output like it ends in a question mark.
“Use it not as a final draft, but as a prompt back to yourself: Is this what I really mean? Is this the most honest or useful way to say it?” he says. “The machine gives you the most probable version of your idea. It’s your job to make it improbable and meaningful.”
So, proceed with caution, but also with curiosity.
And keep a close eye on your cognitive debt.
Slogging through that neural sludge? We’ll help you clean it up.
