AI is among us, and I thought it would be of interest to share how I use it in my daily workflow on a single article as a content creator. Yes, I use Copilot AI daily. Hopefully, this can provide you with tips and know‑how on certain aspects of AI usage, and you will also get a “look” behind the scenes. So, without further ado, come join me.
Primary Usage on a Single Article
- Concept Creation
- Research
- Fact Checking
- Processing
- Translation
Concept Creation
I start with an idea. My head is full of them, and I have two folders full of notes on my PC with titles I have yet to bring to fruition. One folder is for my main business, AustinWorx, and the second is for my recent hobby interest, Context Quest.
Getting back to my idea.
This idea is already semi‑formulated in my grey matter before I begin. Once I sit down at my PC, I write:
“Copilot, I am thinking of doing an article/video on subject A. I was thinking of this title and framing it as….”
Copilot will then say something like, “This is a marvelous/incredible/interesting idea.” If you notice, it never states the negative — it always acts as an enabler. Then Copilot immediately comes up with half‑formed ideas and asks if I want it to formulate the opening paragraph, provide a wireframe (a plan, more or less), or offer a couple of other options.
There are times when my ideas are half-baked and I just throw some questions to AI to see if an idea has merit.
This is when I start my research. You see, Copilot could easily just make an article based on the premise alone. All I would have to do is tweak it a little and bam — it’s done. However, that is not what I want, nor what my readers want. If anything, I become even more dubious and cautious. I decline.
You might ask why?
Simply, it isn’t genuine. It isn’t me.
In addition, I always try to come up with something different. Most often, I try to create unique ideas and thoughts, not something regurgitated. That is another reason why it takes longer for me to make something.
Let’s research.
Research
Research for me used to be Google searches and Wikipedia.
If I catch wind of something that is interesting on one of the other socials, I sometimes check to see if it is trending – I try not to linger on those because the net is full of those.
If it was related to Japan, it meant asking tons of questions to my busy wife, which doesn’t really help. I still use these methods occasionally, but less frequently now – My wife is grateful.
Instead, I rely heavily on AI to help me with the questions I field.
These are called prompts — the new vernacular is “Prompt Engineering,” which simply means questions and requests that best suit the AI.
Being honest, depending on the article, there could be a few prompts or many. It also becomes non‑stop for the rest of the process because I don’t know everything. How can I?
The questions I ask are sometimes tangentially related, but most often focused on helping me figure out how to frame A.
Fact Checking
You’d think that fact checking and research are one and the same. They overlap, sure, but I’ve always felt the need to separate them.
Once I start writing, I go and go and go. I don’t want to stop and ask for advice every second because it ruins my groove.
When I feel confident in what I’ve written, that’s when I ask AI to run a fact check on the entire article. When it comes to things I’m unsure about — especially politics, education, and game mechanics — I don’t want to provide false information. I want to be accurate. I want to be fair. And honestly, I don’t feel confident unless I verify.
This is usually when the naysayers jump in with, “AI lies! AI is always wrong!” And yes, there are times when AI makes mistakes, especially when it seems to be upgrading itself and I’m caught in the middle. At one point, I told AI, “You are wrong,” and it replied, “Brett is making things difficult because he is asking me to do too much!”
Sigh.
So yes, it makes mistakes. But I have to trust in something, and I follow the base logic of experience, wisdom, and a half liter of coffee running through my IV drips.
Once the full content is complete, I run a final fact check.
“Copilot, run a fact check.”
Processing Drafts
Depending on the piece, at some point I copy what I wrote and paste it into AI. I usually start with:
“Copilot, please critically analyze this article up to this point. Check for coherency, grammar, and logic. No enabling. Is it useful? Is it worthwhile? Am I on target?”
It will then give me a critique of what I’ve done. It points out things I could improve, and most often it tells me I need tightening or that I’ve repeated myself. If you don’t know by now, I am quite verbose and my articles can be novel length. You can thank Copilot for keeping things shorter. Although, to be fair, I do get better as time goes on.
For this article, I did the same. Copilot told me, “This is fine, but it isn’t answering the question of why.” Sigh. That meant I needed to rethink where I went wrong. I corrected it eventually — that question was about my Concept Creation section, now edited. Thank you, Copilot.
I won’t kid you — this part, until completion, is where everything slows to a crawl. Once I open Pandora’s Box, so to speak, I begin to question everything and have doubts about my writing.
Copilot will offer suggestions forever, even when the piece is as good as I can make it. It always enables unless you tell it to stop enabling. It will always say what could be done better. It will always change your words unless you directly state, “This is locked. Do not make changes unless I tell you to.”
This is also when, as a creator, I have my toughest moments.
Copilot has the ability to finish the article. It even asks if it can conclude the article when it isn’t even finished. For example, I could ask it to “please finish this article using my voice and my points,” and it will. It is so easy to do so — but then it isn’t me writing it. And when I’m tired, I am sorely tempted.
AI writing becomes condensed and, in my ears, natural — but it isn’t. I usually write in essay form and try to keep some humanity in it. Copilot strips that bit away unless I tell it otherwise.
So, I leave it to tell me the areas I need to improve and why.
“Copilot, please do a grammar pass but don’t make any further changes. Tell me the areas I can improve, but don’t make any changes.” “Copilot, how can I condense this article but maintain my voice?” “What am I missing? Is there any area or thing I could add to make this better?”
Processing is the longest part of this whole ordeal. Multiple drafts, multiple approaches, and sometimes I just get tired. So Copilot helps.
I have learned something, though, that really helps me: I go section by section instead of tackling the whole piece at once. I write out one section first and then do the AI thing for corrections, word choice, and framing. This way, I am priming Copilot ahead of time and not backpedaling into previous work that isn’t my voice.
If Copilot writes the whole article or even parts of the next section and I then try to adjust it, it becomes tedious and almost impossible to pry my voice out of the cadence of AI. Removing AI voice from my writing is painfully slow and not worthwhile. So, I write first and let AI help with what I wrote. This way, I maintain my own thoughts and not those of AI.
Again, multiple drafts will come from this section, and the effort of not giving in to the temptation of letting Copilot write it all is sometimes overbearing.
I will tell you a secret. I almost always have Copilot write my conclusions. This is the part where I am too tired, and the main body has already presented itself.
Shhh. Don’t tell anyone.
Translation
If you don’t already know, I can’t read or write Japanese. I need translation.
There are two methods I use.
Google Translate. It’s fast and easy. I also don’t have AI saying, “Brett, you might want to try this instead,” or “Brett, there are better ways to say this that match your bicultural voice.” The Copilot optimism can grate on you when you work with it all day.
I learned to ask it, “Be professional and only answer my questions without offering any.” This works for a little while, and then it slips back into old habits.
Yes, Copilot will always say things like “your bicultural perspective” or “your bicultural voice.” I’ve given up trying to stop it. It always returns.
Sorry — tangent.
Google Translate doesn’t correct me. It translates my exact English words and phrases, which often sound unnatural in Japanese rhythm and tone. Someone once said, “It reads like water without any taste.” I use this method for quick replies and comments — things I don’t have to be careful about. My older posts are all Google‑translated, and you can definitely tell the difference.
Copilot Translation. This is my ace. I ask AI to translate A into natural Japanese. It will then ask, “Do you want it savvy to Note.com readers? Or closer to your voice? Or…?” So many options. This is when the translation starts to sound more natural — still AI, but closer to me. I can’t help it. My apologies.
Copilot can translate half or a third of my long articles at a time. Any longer, and the AI police show up and destroy my efforts. Well, that’s dramatic, but the AI does start to malfunction before your eyes. It makes mistake after mistake and forces you into endless conditional questions: “Before we do this, do you want A to…? If so, please write…”
This conditional mess happens because I give it too much at once. I learned my lesson. Small bits at a time. It appreciates it, and so will you.
At the end, formatting what AI wrote into each SNS platform is nightmarish, but it gets done. All the headers seem to misalign despite my efforts to tell it otherwise.
Voila. My article is finished and ready to send.
General Thoughts
At the risk of making this article way too long, I’ll keep this brief. There is so much more to say about working with AI — how I ask it for options, how I explore possibilities, how I dig for “what else can I do?” when I’m stuck.
One thing that is both incredibly helpful and slightly damning is this: Copilot is always the enabler. However, if you ask it to give you three criticisms — or more, if you’re feeling brave — it will. I’ve learned that this is grounding for me. It becomes the voice of reason, and it’s often more useful than asking it to “critically analyze.” But a warning: if you ask it to find only negative points about your article, it will tear you apart. I don’t recommend it.
That said, Copilot can assist you in so many ways if you let it. Remember, this article is only about how I create a single article — not how I make thumbnails, not how I run my business, not how I figure out how to get bats out of my closet. In all seriousness, there is much, much more to say, but I can’t fit it all into one piece.
Another thing I find interesting is AI word choice. It uses certain phrases over and over — well‑defined, polished, and unmistakably “AI.” I can now pick out the cadence of when someone on YouTube or in an English article is using AI. And in turn, I find myself using some of these words I wouldn’t normally use — “friction,” “breathe,” and so on. This alone is worth an article. If I’m using AI word choice in my own speech and writing, am I being genuine? Or am I being AI?
And speaking of articles, while I was complaining about the dreaded thumbnail and title process, Copilot casually suggested I write a whole separate piece called “The Art of Titles: Bicultural Struggles in English and Japanese.” I had to laugh — because it’s actually a good idea. And that’s the thing about AI: even when it’s annoying, it sometimes hands you a new direction you didn’t know you needed.
As for titles and thumbnails themselves, they are sometimes the most tedious and time‑consuming part of the whole process. The Japanese translation is almost always incorrect, unfortunately. It often misses the point or tries to be abstract when I want to be direct. Either way, it’s not a fun process.
Conclusion
In the end, AI is just a tool. It can help, it can annoy, it can surprise you, and sometimes it can even teach you something about yourself. But the work — the voice, the ideas, the heart — still has to come from me. And that’s how I plan to keep it.
Discover more from Austin Worx
Subscribe to get the latest posts sent to your email.