I’m not an AI expert. I’m a content marketer who’s been writing online since most people called it blogging.
I didn’t adopt AI as quickly as I took to blogging, but I needed some time to warm up to the biggest potential threat to the American labor force since outsourcing.
Now that I’m here, I’m excited. It’s only been a month, but I’ve been taking notes about the experience as it unfolds.
In this post, I’ll share my initial impressions of content creation with AI. I’ll also pose some follow-up questions for future consideration.
AI is a mirror—what you give is what you get
The depth, clarity, and originality of a GPT’s responses depend largely on what you feed it.
It doesn’t pay to skimp on context or turn your brain off: If your input is half-baked, you can’t expect much from the output.
Take, for example, the very first time I used ChatGPT to help with a content project. I needed to create:
- Landing page copy for an executive dinner event
- An invitation email to a curated list of VIPs
- A “Thank you for coming” note for the post-event goody bag
I wasn’t using AI for content yet—I only reached for ChatGPT because the project specified using AI to help write the “Thank you” note. I didn’t even have an account at the time.
It seems almost comical now, but I didn’t use AI for any other aspect of the project, even as I struggled to tease out fully articulated benefits of attending the dinner.
And because I limited its use to the “Thank you” note (something I could have easily written on my own) I was less than impressed by the experience overall.
I didn’t realize I was the problem. While I put tons of detail into the “Thank you” note prompt, I didn’t do enough scene setting in the beginning. If I’d committed enough to the process to create an account, I probably would have explored further, but I didn’t.
I wasn’t totally open to the experience, and as a result, I didn’t like what I saw in the mirror.
This may be an extreme example, but if you approach AI content creation with resistance, you’ll significantly limit its potential to please.
If I were to approach this project today, I’d feed it tons of context—from information about the brand/event to inspirational copywriting resources. I’d also use it to assist with every asset (except maybe the emails, those were really good).
Artificial intelligence studies everything you share and reflects it back, flaws and all. Thoughtful, nuanced, and self-aware prompts yield the most useful responses.
💡 Key takeaways
- Frontloading your request with context makes all the difference
- Good prompts come from thoughtful process
- Vague or halfhearted prompts = vague or halfhearted output
AI is a synthesizer, but you make the music
AI makes sense of datasets. That means it can help you organize and synthesize information while identifying meaningful patterns.
Knowing this, I asked ChatGPT to draft the oft-dreaded conclusion to a blog post.
You know when your brain is maxed and you’re aware of all the little brilliant pieces that’ll pull the piece together but can’t bridge the gap without a break? That’s where I was.
So I fed it my draft, shared my raw thoughts about what the conclusion should achieve, and asked my new assistant to pull things together.
What I got back was technically accurate but emotionally flat:
I wasn’t happy with either option. They were too long, too meandering, and frankly, a complete bore.
This isn’t to suggest asking was useless. When I returned to the post with fresh eyes, I was able to use both options as a springboard to say what I really wanted to say:
As it turns out, the agony of conclusion writing is, at least for me, an unskippable part of the process, especially when you can’t access the smartest available GPT model.
Even then, we all recognize shameless AI writing when we see it—and it’s not just the em dash. There’s a flatness and a “cuteness” to the way it phrases and frames ideas that feels… well, artificial.
Conclusions aside, the process of planning this post made ChatGPT’s synthesizing and organization skills shine.
First I shared my raw notes in exchange for feedback and an outline, then I provided feedback to receive an even better outline.
Given my feedback, ChatGPT helped me to group similar points into broader insights, a process that typically takes some time.
I reviewed the new outline in a Google doc, slotting both my raw ideas and the initial outline notes into their respective sections. A virtuous circle.
I probably could have plugged this into ChatGPT for more refining, but I thought better of it.
Writing is a thinking exercise, and allowing AI to handle too much heavy lifting robs you of that reward and disconnects you from the process.
💡Key takeaways
- Useful for momentum, not necessarily mastery
- Don’t expect magic without some elbow grease
- “Does this sound like ChatGPT?” is now part of the writing process
AI is an extra (non-sentient) brain in the room
One of the most exciting aspects of AI content creation is its ability to surprise you. When used with intention, it becomes more than a tool.
It’s a brainstorming partner, an idea generator, and pattern finder—an extra brain in the room that reflects more than you might expect.
When I requested a conclusion for my backlinks post, for example, part of its output included copy that became the tagline for my blog. That same conclusion prompt also led me to ask a question that surfaced an idea for a future post.
Would I have arrived at these ideas anyway? Probably—but I’m not mad AI sparked them.
Realizing its strength as an idea generator and momentum-starter, I also used ChatGPT to help rewrite homepage copy, something I’d been sitting on for longer than I care to admit.
Sometimes the surprise is validation. More than once, it’s shared a framework, phrasing, or idea I’d already been toying with but hadn’t mentioned.
That kind of unprompted reassurance is encouraging when you’re working solo, especially on longer-term or loosely defined projects.
It also makes admin tasks and troubleshooting more manageable. From one interface I can solve WordPress issues, refine headlines, and talk through metadata strategy. For its convenience and versatility alone, using AI for content creation is a win.
The big shift was learning when to let AI lead, at least a little. Not in a vending-machine way where you enter a query and expect brilliance in return, but in a “yes, and…” way, where you remain curious and open.
💡Key takeaways
- AI can surprise you, but only if you let it
- Idea generation is its quiet superpower—don’t overlook it
- Thoughtful, structured asks = thoughtful, structured responses
- Treat it like a collaborator, not a vending machine
- It’s more than a ho-hum writer—tap it for structure, support, and strategy
Image creation is tricky but exciting
Using generative AI for visual content is like working with a designer who doesn’t read briefs.
You can absolutely land on something striking and unique, but it’ll likely take multiple tries, several tools, and some example images to get there.
Some tools, like Claude, can create futuristic interactive visuals using HTML and CSS, an inspiring discovery I wasn’t expecting.
Unfortunately, the results are often unpredictable. I had to ask for a basic file format three times before Claude got it right—then it abruptly cut me off mid-convo.
Others, like Canva’s AI tools, offer better consistency and formatting, but even then, following basic parameters doesn’t guarantee a usable output.
Instructions get ignored. Styles shift unexpectedly. Results vary, even with nearly identical inputs.
While creating images for this post, I used identical prompts across AI tools, and it was fascinating to see the variations in output—sometimes from the very same tool.
Still, the process of experimentation is valuable. I’ve prompted my way to visual assets I couldn’t use (yet) but loved enough to save anyway.
AI images feel like the future, but they’re not quite there yet—at least not for this non-paying content creator.
💡 Key takeaways
- Embrace the chaos, save your prompts, and get examples of visuals you like
- If you don’t want text or numbers in the image, say so (and hope it listens)
- Learn how to talk about what you want for better input/better output
- Paywalls are more impactful here than they are for written content—plan accordingly
Rewriting content creation with AI
How many of these AI tools will still be free in six months? Will free tools eventually become like YouTube downloaders and torrent sites? Will paywalls choke image generation altogether?
I don’t have the answers. But I do have a few takeaways.
AI doesn’t replace human creativity, it reflects it. The more you bring to the table, the more you’ll get back.
For now, this means I prompt more. I plan more. I revise with more intention.
I’m already using less em dashes.
