The Question Worth Asking
The real question isn't "Can AI write this?" It's "Should this be written by AI?"
I don't publish any blog post that's 100% AI-generated, even when the AI output is grammatically perfect and technically accurate. There's something lost when a machine does all the thinking. Readers can feel it.
This isn't gatekeeping. It's a recognition that different content has different requirements. A product specification document can be fully AI. A personal essay should not.
AI Strengths in Content (Real Ones)
Brainstorming and structure: AI is fast at generating options. "Give me 10 possible angles for a blog post about Make.com automation"—AI does that in seconds and the ideas are reasonable.
Boilerplate and templated sections: Introductions explaining why something matters, explanations of basic concepts, calls to action—these are semi-standardized. AI handles them well.
Rapid iteration: Need 5 versions of a headline? 3 different email subject lines? AI is faster than your brain.
Scale of summary work: Taking existing long-form content and converting it (article to tweet, article to social post, long email to short email). Format conversion at scale.
Human Strengths That Still Matter
Original insight: You've built something, failed at something, learned something. That lived experience is irreplaceable. AI can't write "here's what I tried that didn't work and why." It can write "common mistakes include..." which is different.
Opinion and conviction: AI hedges. It says "some people believe" or "it's arguable that." Readers come for your takes, not AI's careful neutrality.
Narrative voice: AI can do voice for a section. Can it sustain your voice—your way of phrasing things, your rhythm, your personality—for 2000 words? Not yet.
Research and judgment: You know which studies matter, which "research" is BS, which context is important. AI hallucinates citations and misses nuance.
Cultural timing: Knowing what's happening in your industry right now, what your audience cares about, what's topical vs. evergreen. That's judgment, not pattern matching.
The Hybrid Approach That Actually Works
Here's what I do:
- Idea and outline (me): I decide what's worth writing. I outline the core insight and key points.
- First draft (AI): Claude or ChatGPT writes sections based on my outline. This is 40-50% of the work.
- Edit for voice and insight (me): I read the draft and rewrite anything that sounds like AI. I add examples from my experience. This is 40-50% of the work.
- Polish and fact-check (tools): Grammarly for editing, manual fact-check for claims.
The result is something that's faster than writing from scratch but unmistakably my work.
Where I Draw the Line
I would not publish:
- A blog post generated entirely by AI and lightly edited. Too much sounds like an AI.
- Personal or opinion content without significant human rewriting. AI will make it sound generic.
- Technical tutorials or "how-tos" without verifying every step actually works.
- Anything with claims or statistics I haven't fact-checked personally.
I would publish (with appropriate editing):
- Boilerplate sections or introductions written by AI.
- Social media posts generated from existing blog content.
- Product documentation or spec sheets.
- Summaries of my own longer work.
The Authenticity Problem
Readers can tell. Not always consciously, but there's a difference in feel between writing done by a thinking human and writing done by a language model trained on the internet.
AI prose is smooth, grammatically perfect, and emotionally cold. It's like reading a manual written by someone who's never used the thing they're describing.
Human prose has friction, personality, mistakes, and moments of clarity that come from actual thought.
If your content's value is speed and volume, maybe AI-generated is fine. But if your content's value is authenticity and trust, you need the human in there.
Building for the Future
AI will get better at reasoning, voice, and sustaining personality. It might get good enough that some readers can't tell the difference. That's coming.
But it won't replace the need for human judgment about what's true, what matters, and what's worth saying. Those are human decisions.
The future of content creation is probably: AI as a collaborator that amplifies your output and speeds up the mechanical parts, while you focus on the thinking and judgment parts.
Practical Takeaways
Use AI for: Structure, rapid iteration, format conversion, boilerplate.
Keep human for: Insight, opinion, voice, and verification.
The content that will win in the future isn't purely AI or purely human. It's the stuff where humans use AI as a tool to amplify their thinking, not replace it.