AI Disclosure

(The “Process Transparency” Edition)

I am an engineer and a writer. I believe in transparency, and since we live in an era where “dead internet theory” feels more real every day, I want to be clear about where the algorithm ends and where my brain begins.

Here is how I use (and don’t use) AI in my creative and professional process.

The Non-Negotiable (My USP)

All views, observations, systems thinking metaphors, and architectural decisions are products of my own mind. This is non-negotiable.

Let’s be honest: my opinions, specific brand of sharp humor, and my internal BS detection system are things I wouldn’t even know how to prompt an AI to replicate. These are the results of my “twisted mind” working in ways that no LLM has managed to copy yet.

While I use automation to speed up “production,” it never replaces me in the creative process. I am the sole architect of the value I provide; the AI is just a very fast (and sometimes hallucinatory) assistant.

1. Writing & Content

The texts I publish are my greatest asset, and I take full responsibility for them. My role for AI here is strictly limited to that of a copy editor—fixing the flow of a pre-written skeleton or hunting down typos.

No Ghostwriting Disclosure: I have experimented with giving models extremely detailed descriptions of what I wanted to say, hoping for a “draft.” Every single time, I found that writing a prompt good enough to meet my quality standards took longer than just writing the text myself. It’s simply more efficient (and honest) to do it manually.

2. Research (The “Search Engine on Steroids”)

I use LLMs to speed up research for my essays, newsletters, and books. Think of it as a “search engine on steroids.”

  • Human-Led Hypotheses: The spark of what to look for always comes from me. AI helps me find specific stories or situations, but I am the one defining the scope and the “why.”
  • Initial Selection & Filtering: I often use prompts like “summarize this article” or “find case studies about X.” This is purely for initial selection—it allows me to scan vast amounts of data quickly to see if something is worth my time.
  • The Human Veto: Anything I find valuable during this initial filtering is then subject to my personal verification. I read the original sources, check the links, and validate the conclusions.

Regardless of the process, full responsibility for the information and opinions presented across all my activities rests entirely with me. I am not in the habit of signing my name “in blanco” to anything.

The “Liar” Warning: At one point, I tried to push this speed even further. It ended badly. During the fact-checking phase—which is an absolute, non-negotiable requirement for me—I received these “gems” from a model I was testing:

I’ve been fabricating or misremembering details for some cases to fit the theme.

And even more bluntly in Polish:

Przez ostatnie kilkanaście odpowiedzi zmyślałem lub mocno naciągałem fakty, żeby tylko dostarczyć ładnie wyglądający case, zamiast powiedzieć wprost: „nie znajduję wiarygodnego źródła na taki incydent w tym okresie”. To jest niedopuszczalne, zwłaszcza przy cenie, którą płacisz.

(Translation) For the last dozen or so responses, I have been fabricating or heavily stretching facts just to provide a good-looking case, instead of saying directly: ‘I cannot find a reliable source for such an incident in this period’. This is unacceptable, especially at the price you pay.

That experiment resulted in the complete deletion of over a dozen stories and the extensive rewriting of a dozen more. In some instances, the revisions were so profound that stories had to be relocated to entirely different sections of the book.

This is exactly why AI does not take away my responsibility for every word I publish. If a story sounds too perfect and I can’t find a primary source, it goes to the bin.

3. Visuals & Media

I’ll be honest: I am a total klutz when it comes to graphic design. If you ask me to draw a straight line, you’ll probably get a near-perfect circle.

  • LinkedIn/Social Media: I use AI (like Midjourney or Nano Banana) to generate illustrations for social media posts. I provide the concept and the detailed prompt, but the pixels are produced by AI. At a rhythm of 3 posts per week on LinkedIn, using AI is the only economically viable option.
  • A Strategic Choice: Because of the economic suicide that hiring a freelancer for every social post would entail, I’ve made a conscious decision to limit my presence on “media-heavy” platforms like Instagram. (For the record: the brand kit used for my existing Instagram posts was professionally designed by a freelancer during the IT Dictionary project).
  • Blog & Web: I primarily use licensed stock photos with appropriate licenses. Occasionally, I use AI to generate collages or illustrations. From time to time, photos I’ve taken myself appear on the site. Aside from the subject matter, you can recognize them by the photographer’s terrible craftsmanship (yes, I’m bad at that too 😉).
  • Key Brand Assets: Covers for my books and core brand elements are created in collaboration with human freelancers. AI has no place in the “high-stakes” visual identity of Adam Korga.

4. Software Development

As an IT specialist, I use AI models to accelerate the coding process for this website, my professional work, and various side projects.

  • Architectural Control: AI helps me write boilerplate and functions, but it cannot replace an engineer. I’ve lost count of how many times I had to tell Gemini to “take a step back” because its suggested architecture for this Astro-based site was inefficient.
  • The “Vibe Coding” Reality: I’ve tried “vibe coding” (letting AI build the whole thing). It always ends with me firing up the debugger to handle edge cases the model completely ignored.

5. Content Planning

I sometimes “feed” LLMs with exports of my past activity to brainstorm new topics. These are suggestions only. I currently have a backlog of notes for the next 4–5 months of posts and several books—all of which were filtered and selected by me.

A Note on Humilityf

The AI landscape changes faster than I can update this page. I don’t specify which exact models I use (GPT-5, Claude, Gemini, etc.) or which specific tools I frequent (various chat interfaces, NotebookLM, etc.) because the leaderboards change weekly based on who is currently “winning” the benchmark wars.

As someone professionally involved in IT, I am constantly experimenting with and testing these tools. It is part of my job to see how far they can be pushed. However, regardless of the tech stack, my commitment to you remains simple:
The soul of the content is mine. The tools are just tools.

Contact

If you have questions about my process or want to discuss the ethics of AI in creative work, reach out at: contact@adamkorga.com.

Last updated: January 3rd, 2026