Thoughts and ideas by Brandon Kowallis processing and editing and co-writing by Claude
The moment most people learn that something was written with the help of AI, they stop reading. And I totally understand that reflex. I have it too. It’s not so much that the writing isn’t grammatically correct or well structured, or that the ideas aren’t interesting. It’s the fact that very little human thought, creativity, or effort went into it. And if the writer doesn’t care enough about the topic they are writing about to spend time thinking through the topic or verifying the accuracy of the information they are sharing, why should I spend any time bothering to read it.
What most people are reacting to, however, isn’t AI it’s laziness and inauthenticity. Someone types a few careless prompts, takes whatever the machine hands back, and posts it without reading it closely or questioning whether or not it captures what they themselves are thinking, because in most cases they aren’t thinking at all. Of course nobody wants to read that! The person who put it out there barely wanted to read it themselves. The problem is that in most cases we don’t know what level of effort was put into it, so we just assume that if AI was involved the creator was lazy.
But I believe that is a serious mistake. Artificial intelligence can be a powerful communication tool for those who actually care about sharing their thoughts and ideas. It can take brilliant ideas from a person who is a poor writer or communicator and structure those so that others can easily understand them and grow from their knowledge. It can help creative, deep thinkers who are good writers finalize their ideas into written form much faster, increase their output and our access to their helpful, new ideas. But this only happens when AI is used as a collaborator and not the author.
So allow me to pull back the curtain and show you one way this can be done, using a real example: an essay I recently wrote and pitched to The Chronicle of Higher Education and Inside Higher Ed, arguing that we should stop policing students who use AI and start teaching them to use it well. It is a piece I care deeply about and a piece I built together with AI.
But first, one honest confession about myself.
The problem with writing alone
Writing is a solitary act. I love it for that, but what I don’t love is the impact it has on my ability to articulate things verbally. The more I write in silence, the sharper I get on the page and the duller I get out loud. I get the impression that I’m not alone. Many people nowadays spend so much time emailing, texting, social media-ing, and entertaining ourselves with screens, that clear, concise verbal communication doesn’t seem to be as regular as it used to be.
So when I set out to write the AI essay, I decided not to open a blank document. I opened my mouth. I turned on speech-to-text and I talked. I rambled. I said the same idea four different ways. In twenty minutes of talking I found I can get down thoughts that would have taken me hours to coax onto the page, and I get the added benefit of practicing the very skill that solitary writing erodes — the ability to better connect my thoughts with my mouth and become fully present with the words I am saying.
At the end of that process, what I had was not an essay; it was a big pile of construction material. Thousands of words of me circling the same handful of convictions—that AI is a tidal wave, not a graphing calculator; that we are spending more energy catching students than teaching them; that there is writing you do for the writing and writing you do to deliver something else. Real ideas, all of them mine, buried in a pile of verbal material — a bulky data set of ideas.
Turning the pile into a first draft
I could spend hours building out the perfect structure for that first draft. And I do at times, for personal works where the joy of creation and personal achievement is more important than sharing my ideas as quickly as possible. A person might use a lever if they are trying to move a heavy object for a construction project, but if they are moving the heavy object just to see how heavy and object they can lift without any help they would not use that lever, because it would be cheating. There is a time to laugh and a time to cry, a time to use AI and a time to not use AI.
What is AI really good at and really fast at? Organizing and processing data. What have I just created? A large set of data. So, for the next step in the process I take that mess and I ask AI to do what it does best: finding the house hidden inside the pile of material. But I don’t do it with a lazy prompt. I build out long, specific, carefully designed prompts that tell the system how I want it to think, what rules to follow, what structure to build toward, and to write it in my voice.
Anthropic’s Claude is the best tool I’ve found for this, because it lets me keep a database of my own writing: essays I wrote entirely by hand without any AI assistance. That library is my voiceprint. When I ask it to organize my rambling into a first draft, I am not asking it to write like a generic robot. I am asking it to write like me, using me as the reference. The draft that comes back is not the final essay and it is not meant to be. It’s scaffolding. It takes my pile of material, lays a foundation and frames the house under my directions — an opening, a middle, a conclusion — in a fraction of the time it would take me to construct that myself.
And sometimes the draft is not even close to what I am satisfied with. That is totally fine; part of the process. I rewrite the prompt, add guidance, point it in a better direction, and see what comes back. I keep guiding it until the foundation and structure is good enough to actually work with. What used to be the most agonizing part of writing — staring at the blank page, wrestling a shapeless mass of thoughts into some kind of order — now happens fast. And that fast first draft does something I never expected: it sparks creativity. Seeing my own ideas laid out in a structure gives me something to push against, build on, and argue with.
Becoming my own worst critic
This is where the real writing happens. In the AI essay, this is the stage where my initial random ideas began to coalesce into into a solid argument to back up my random opinions. But it is also the step most people skip, and it is the one that matters most.
Once I have that first draft, I throw out 80% of it and rewrite the entire thing myself. Because if you read an AI first draft quickly and without much thought it looks great. But if you read it the way a grumpy critic would, someone whose whole purpose in life is to tear your argument, grammar, and clarity apart, you find that it is nowhere near as polished as it needs to be. So that is exactly how I read it. I sit down as the enemy of my own work. I hunt for every weak claim, every place a skeptical reader would say “prove it,” every argument that sounds nice but rests on a shaky foundation. Then I go find the answers, rethink and rewrite it, in my own words, addressing the objections head-on.
The long, messy, wonderful middle
When I rewrite this way, something predictable happens: the essay gets better (duh) . . . and it gets much, much longer. What started as a tidy, hollow draft swells into a sprawling, overstuffed document that is now eighty or ninety percent my own original thinking. The structure is mine. The arguments are mine. The examples, the turns of phrase, the convictions — mine. It is also, frankly, a bloated mess. But it is the right mess. It contains everything the essay needs to say. It just says all of it twice.
Polishing—and the honest line between tool and voice
So now I hand it back to the machine, but for a completely different job. This time I am not asking it to think. I am asking it to trim. Tighten the grammar, cut the redundancies, find the three sentences doing the work of one and collapse them. This is work AI does quickly and well, and it is work that would take me hours of tedious pruning and analysis to do by hand. The ideas do not change. The voice does not change. The essay just gets lighter and easier to read.
And then comes the last step, which is entirely mine again. I go back through the whole thing and put my fingerprints all over it. I rewrite the bits where the machine’s phrasing is competent but sterile. I drop in the sarcasm, the sharper metaphors, the lines that only I would write. I make sure that when you read the final piece, you are reading me—not a smoothed-over average of me. Because a tool can help you build the house, but you should be the one who decides where the light comes in and the style of the decor.
When I use AI—and when I would never
I want to be as clear as I can be about this, lest you believe that Brandon has gone off the deep end of the AI swimming pool.
When I am writing for pleasure—writing to work through what I actually believe, to exercise my imagination, to make something I can look back on years from now and feel proud that I made it—I never touch AI. Not once. That writing is where the joy lives and where I grow, and handing it to a machine would be like paying someone to lift weights for me and wondering why I never got stronger. That space is sacred, and I guard it.
But a great deal of writing is not that. A great deal of writing is a delivery mechanism—a container whose only job is to carry an idea from my head into yours as clearly and quickly as possible. When the point is the message and not the making, AI becomes a tool that lets me build that faster, while still maintaining high quality and readability. And here is the part the critics miss: doing this does not take writing away from me. It gives me more of it. Every hour I do not spend manufacturing a document whose only job was to transmit information is an hour I get back for the writing I love to do, that brings me joy, and helps me to grow.
That is the real reason I do it. I have a head full of thoughts about artificial intelligence and where it is taking us, and I genuinely want to share them, because I think we all need to be thinking hard about this together. But I also have stories and ideas rattling around in me that I much would rather spend my time on, and there are not enough hours in the world for both. Using AI to help me deliver the first kind of writing is how I buy back the time for things that matter more to me and fulfill me. It is not a shortcut around the work. It is a way of spending my finite writing energy on what matters most.
So read it however you want
You are, of course, free to find out that this essay—and the one it describes—were written in collaboration with AI and dismiss them on the spot. That is your right. But I would gently suggest that the question worth asking is not whether a machine was in the room. The question is whether a human being actually did the thinking. In my case, I promise you I did. I talked until I found what I believed. I tore my own arguments apart and rebuilt them. I fought for every line that sounds like me. The machine helped me build it. It did not tell me what to build.
And that, I would argue, is exactly what we ought to be teaching our students to do—not to fear the tool, and not to surrender to it, but to stay firmly in the driver’s seat while it does the heavy lifting. Which is the whole argument of the essay this one was written to explain. I just thought you deserved to see how the sausage was made.
