Part 2 – Before You Begin: KSA – Knowledge, Skills, and Abilities

This is Part 2 of a series by B.A. Bellec on using AI to edit his novel Cobalt Lungs. Start with the [Series Introduction].

Why a decade of beta reads and failed drafts, not raw intelligence, is what made AI useful to award-winning author B.A. Bellec.

I am going to brag a little. Not everyone writes an award-winning novel. Not everyone makes music tied to their books.  Not everyone runs a YouTube channel. Because this is my 4th novel and I already have close to 20 songs and music videos done, I have a very unique set of creative-level KSA (knowledge, skills, and abilities). Then on the business side, I know automation, process, and highly technical software. Again, those are very unique, tech-heavy KSAs in my toolkit. My unique blend is a huge reason I was able to pull this off and document the entire process. My goals were clear, and I knew I was walking a rare and controversial path; that feeling is actually a good thing. It means I am aware and doing tasks with intent instead of randomly scattering ideas around and hoping for the best.

I have been through multiple structural edits with professional editors for my prior novels. I have had probably 20 total beta readers across all my books. I have beta-read close to a dozen unpublished manuscripts for other authors over the last 3 years. I have worked with amazing artists in both the comic book and music industries. These all added up to a huge inventory of skills I could draw from. I consider my project a success, not because I used ChatGPT to help, but rather when I started using GPT, I set a ridiculous bar: this was going to be my best project to date. Period. Full stop. If I am not satisfied it is above the level of everything I did before, I rework the scene, the moment, the character, the lore, the song. That bar is high, and I could only set it because of the vast skills I acquired over the last decade.

AI is not a replacement for learning how to create art. If you try to just jump to the finish line, it won’t work like you want it to. You still need to learn the fundamentals of creating great art. There is always more than one path, but I highly recommend you use beta readers, consider a structural edit, and do a copy edit. These are all important steps in the book creation process, and if you skip them on your early publications, anything you try to do in GPT will be like the blind leading the blind. You won’t know what to ask (prompt) because you have never done it yourself. You won’t know what to look for on the other end to make the corrections and adjust or calibrate. You have to clearly identify your project phase with GPT at all times to make sure you are getting the correct feedback.

I left an important topic out of the prior article. Education. Mixing generative AI with education also worries me. There are reports of generative AI being used by students to answer questions or write full papers. Effectively, the students are skipping the learning exercise. The point of education is not to get good grades, graduate, or finish all the work. It’s to learn. To analyze. To think. Many assignments are designed to build skills, and if the only skill you build is prompt writing, you’re actually handcuffing yourself because you won’t be able to audit the output later when you’re by yourself. Lots of writing in school is to teach students how to edit and audit their own work, or how to seek feedback. That is effectively the core of the learning exercise.

The education system has lots going on besides just the use of AI, and I get that there are other bigger and more complex problems like funding and classroom sizes, but skipping the learning exercises altogether is not the solution. I can see use cases for AI as a tool near the end of the education pathway to help brainstorm and organize big thoughts, but if you start the usage too early, you skip all the critical analytical skills that will make generative AI truly powerful. How are you going to defend and create your work if you don’t even understand the fundamentals driving the output?

I view myself as a naturally gifted learner. I have always been able to pick up things fast, but language and communication were never subjects I excelled at. You could say I am a late bloomer when it comes to my ability to grasp language, and much of my grasp actually comes from my systematic approach to tasks. I got good at writing because I already knew how to compartmentalize and systemize the things I was weak at from a lifetime of learning and working.

I think the number one skill overall is actually experience. Intelligence got me started. Experience made me good. They are not the same thing. You need both, but experience starts to overtake intelligence pretty quickly as you progress through your career or any big project. Intelligence helps you understand or complete a lesson. Experience teaches you why the lesson matters.

My strengths are learning quickly, pattern recognition, retention, and systems thinking. That’s intelligence. Those things absolutely helped. It’s why I have these much larger skills like Finance, automation, coding, music, marketing, and writing, because they bloomed from my core strength, but intelligence alone didn’t solve Cobalt Lungs. Experience did. I had to put everything together and know exactly what to do, and when.

Honestly, when I started my ChatGPT journey, I was already intelligent enough to use AI, but I wasn’t experienced enough with AI. The version of me from Month 1 and the version of me from Month 3 are dramatically different AI users. I didn’t get all that much more intelligent. I have mostly the same storytelling skills as I did two months ago, but my experience has changed. I learned when to trust AI, when to challenge AI, when to brainstorm, when to stop, when to edit, when to ignore feedback, and when to push harder. Those are all instinctive decisions that come from my last decade of experiences interweaving with my current tasks.

My first novel won an award. That’s impressive, but I can tell you this. I am not the same person who wrote that novel. Not even close. Not because I’m smarter. I don’t feel any smarter. I barely feel different. I have done a lot more things since then. I’ve written four novels. I’ve edited thousands of pages. I’ve worked with fantastic collaborators. I’ve received reviews, good and bad. I’ve succeeded. I’ve failed. I’ve repeated the cycle. That’s experience. That’s why I was able to use ChatGPT so well. Try not to confuse being talented with being finished. A gifted beginner is still a beginner. A brilliant first draft is still a first draft. A smart person still needs repetition and failure. You have to do the work. You have to make mistakes. You have to push through the uncomfortable phase where you realize you don’t have the answer yet and your intelligence isn’t enough, because eventually experience starts stacking on top of intelligence, and that’s where real expertise comes from.

Outlining, brainstorming, and structuring are perhaps the most important writing skills you should have to help with communicating your idea, and I got most of these skills from my day job, not from writing books. If you are working with an editor or with GPT, you need to be able to show and communicate what isn’t there yet as much as what is there. If you have strong outlines, useful brainstorms, and a well-scoped project, any partner you bring on is being set up for success because you can communicate the phase and expectation. It’s not just about the manuscript. It is about analyzing your own work. Take a day every week or two and focus on the intangibles of your project. These outside-the-manuscript communication pieces will pay the time you invest back tenfold once you start partnering with other creatives or start using complicated software.

I will say that the biggest mistake or knowledge gap I had when I started sharing with GPT was that I had no idea what the GPT system limits were. I uploaded 13 chapters with strong outlines for about 7 more, and then a brainstorm for how I thought I might end it. I was under the impression it would read and remember everything, but that isn’t how GPT works. It has amazing recent memory, but it will not remember everything you did a few days ago. I would start progressing my brainstorms and outlines into new chapters, and I would notice suggestions from GPT that include characters from later in the book, or characters that didn’t even exist in my story yet, and came from a brainstorming doc.

I had to stop and backtrack to see where the idea even came from, and often it was GPT getting confused about which document should take priority. I had to start telling GPT which things were manuscript locked and which things were still in development. Which chapters and characters were allowed to move around and be reused, and which characters couldn’t have early story changes? GPT would have butchered the story if I didn’t raise the flag, backtrack, and fix the error. I also found that working on little 3-4 chapter sections was how GPT did best. It was really good and remembered the last few chapters and brought forward those ideas, but it was not good at remembering chapter 1 when working on chapter 25. If Chapter 1 mattered to what we were about to brainstorm, I would upload it or summarize it to make sure GPT has the fresh facts in its memory.

It took me about a month to learn how GPT worked and communicate my story’s big picture. That is not to say the first month was a waste. Not at all. I picked the brainstorming that improved and wrote out some of the best ideas into scenes and chapters. The manuscript grew a solid 10K words; it wasn’t the first draft done, but it was starting to progress closer to my target, and my ending was coming into focus. My first month was largely spent learning how to properly communicate with ChatGPT and building workflow processes on my end. It was only after I understood some of the limitations and had the right workarounds that I could start moving forward effectively.

ABOUT THE AUTHOR

B.A. Bellec is an award-winning Canadian novelist, songwriter, and music video producer whose work blends cinematic storytelling, emotional atmosphere, and deeply human science fiction. Drawing from an education in film and a multidisciplinary creative process that merges music, prose, and visual symbolism, his stories explore themes of survival, memory, grief, identity, and the search for meaning inside collapsing systems.

Known for emotionally immersive worlds, recurring symbolic imagery, and a style that balances intimate character work with large-scale thematic ambition, Bellec creates stories that feel less like traditional narratives and more like lived emotional experiences. His current project, Cobalt Lungs, expands beyond the page into a companion concept album, combining literary fiction, atmospheric worldbuilding, and music into a unified creative universe.

COBALT LUNGS

Station Eleven meets The Road with the psychological containment and systems tension of Silo.

A deadly airborne spore has transformed the world into a collapsing maze of quarantines, sealed cities, and desperate survivors struggling to adapt. In the ruins of Vancouver, musician Maddy fights to hold onto her humanity inside an overcrowded refugee shelter while scientist Ethan isolates himself nearby, deep underground in a self-sustaining silo where routine has replaced emotion. As society fractures and survival becomes increasingly brutal, the two are pulled toward each other in ways neither of them fully understands. Atmospheric, intimate, and emotionally charged, Cobalt Lungs is a post-apocalyptic story about isolation, survival, and the dangerous hope that can still burn at the end of the world.

LEARN MORE