How to implement AI strategically, without losing your soul or mind
Picture this: you're reading a Substack you've loved for months. It has made you feel seen and has clarified thought patterns and personality quirks you’ve always wondered about yourself. Then you hit a line that's clearly a leftover prompt pasted into the copy, and your heart drops. You’d thought that you’d been in conversation with an expert who has discovered pieces of your soul. Instead, it becomes clear why this Substack author sent out 4, sometimes 5 new pieces a week: it was largely AI-written, and you can now detect the patterns easily. You unsubscribe right away, out of instinct, not snobbery. The soul has left the room.
The seduction
Look, I get it: AI feels like it cures writer's block and gives you back time. It can catch the typos and the contradictions, create that through line you just can’t quite word perfectly, and add that final thought that you may have taken weeks to get to. Sometimes I use Claude to help me bullet out my own voice-noted thoughts in the early stages of writing these very pieces. But the trade-off is real: you gain speed and lose the struggle, and the struggle is where the meaning gets made.
Writing can also be how we find out what we think and take readers on a journey. So if you’re writing something personal, not for the purpose of expertise dispersal, the actual act of following a mental thought with a pen can uncover so much. This is why journaling is so impactful for some. When you hand that off, you give away the self-discovery, not just the labor.
Why we can tell
Noticing AI-written content is more than just too many em dashes and ultra-wordy examples in triplicate. Readers are pattern detectors; we sense the smoothness, the tidy symmetry, and the lack of risk even when we can't name them. Reading someone’s inner world, someone’s expertise, or someone’s run-on sentence carries a human connection. When I pick up a book, it is often the writing style of the author that keeps me engaged or not. AI gives you data over depth: the statistically likely sentence. A soul gives you the one only you would write.
Research backs this up, too. In the now well known MIT study, human graders described many of the AI-written essays as generic and soulless, with standard ideas and repetitive language. AI also flattens voices across people. One study found LLM access boosted individual story creativity but significantly homogenized output across writers. Everyone gets a little better and everyone sounds the same. So, when you read an article, an essay, a cover letter, or an email and all you get is a word salad with no dressing that sounds somewhat smart, it is likely that ChatGPT is afoot.
The expert problem and what it costs us
If we cannot be trusted as experts to put our own words on the page, who can? As we venture into the Wild, Wild West of this technology, we'll find a scarcity of true perspective and experience, and a flood of flat, technological prose masquerading as expertise. The pendulum will swing to authenticity, and where will we be when we have already outsourced it for a bit more speed and ease? Authorship is a form of attunement. When the writer isn't present, the reader can't feel met.
One of the scarier parts: early research suggests that outsourcing our authorship and ideas to a well-fed machine may be reshaping how we think. Social media and quick-clip culture have already done a number on attention spans (when is the last time you read an entire book?). In “Your Brain on ChatGPT” (Kosmyna et al., 2025), MIT's Media Lab found that brain connectivity scaled down with the amount of external support. Writing unaided produced the strongest brain activity and the widest networks, search engines were intermediate, and ChatGPT use produced the weakest. Eighty-three percent of the ChatGPT users couldn't quote from essays they had just written. The authors call this “cognitive debt.”
The other side: AI as strategist, not ghostwriter
So far, this all sounds like a big trouncing of AI culture. As a fractional Chief of Staff whose job is partially to make things more seamless for the companies and leaders I partner with, ignoring or demonizing AI wouldn’t make a lot of sense. The problem isn't AI; the problem is handing it the wrong job. AI is very good at structure, sequencing, pattern-spotting, pressure-testing, and building systems. It's poor at being you. Let AI build the architecture and keep the voice for yourself. The architecture is the pathway and the strategy. The voice is the knowing, and the knowing is how we all fly.
How I'm doing it
When I first recognized what was happening with AI, I was hit with fear. There is so much unknown and so much power and magnitude behind this technology. As always, there is room for so much good and so much recklessness. We have seen that recklessness already. As I started to study and understand what AI can do, I vowed to learn as much as I could while also retaining my own voice, soul, and ethics.
In understanding AI, I used AI itself. I asked my AI partner for a 30-day plan to learn how to build agents, templates, and models to enhance my Chief of Staff work. I wrote my own operating system, learned how to structure a prompt, set up Projects, and built verification habits. I pressure-tested my own strategies as my own first client. I made repeatable workflows, turning meeting notes into action items, building weekly leadership plans, and mapping strategy. I learned how to customize this for clients, based on the work they are doing. This is the rote work that eats your hours and the strategies I can leave you with that will transform your operations long after I am gone.
The rules that protect the soul
Decide, then draft: AI should be used to structure and analyze your own thoughts and decision-making, not do it for you. Ask yourself, did the final implementation course come from me or a machine?
Find your blind spots: Use AI to argue against your plan and run a pre-mortem. That strengthens your thinking instead of replacing it, and it can help you find your own blind spots, something absolutely everyone has.
Verify what carries weight: Never accept the output on faith. Over-trusting AI is exactly when people stop thinking and pick up mental crutches. Do your own research and keep training yourself to trust your gut. Mission-driven work, especially, should rely on this first.
Keep the pen in hand: Bullets and scaffolds from AI are fine. The prose should stay yours, ideally dictated, because your voice is the most powerful thing you own.
With these strategic repetitions, heart-and-soul led rules, and constantly keeping a finger on the pulse of the good, bad, and ugly of AI, I am able to deploy major time-saving strategies across a company, as well as train others. AI does the scaffolding, the repetition, and the first-pass analysis, but the judgment, relationships, and voice stay with the human, as they should. And believe me: it’s the emotional intelligence and soul where the real progress lies.
Sources:
