The Power Tool vs. The Factory Line: AI and the Creative Works

Published on Aug 01 2026

I wrote a book. Initially, I decided to try the traditional path first and attempted to get published through literary agents.

Enter the submission portals of the creative industry. I'm going to reference a specific clause from an agent's submission form in this post. I want to be clear that this isn't an attack on that specific person or agency. Rather, this style of clause can be found across the entire creative sector today, from literary agencies to indie game publishers, art contests, and production studios. While the example is specific, it is indicative of a systemic trend across every artistic field.

A Slippery Slope: Mandatory AI Disclosure

I've seen variation after variation of this style of AI disclosure, but this one in particular stood out to me:

We stand against the use of AI to generate creative work. If AI was used for research, brainstorming, outlining, drafting, or any other task related to preparing submission materials, please disclose how it was used and to what extent.

0/1000 words to explain yourself in a text box

I find this sort of mandate out of touch, and frankly, unrealistic for creators to accurately answer with every passing year. AI development is rapid, and its integration into software is becoming ubiquitous whether we like it or not.

I'm writing this post in markdown, in an IDE that automatically attempts to guess my next words and lets me hit TAB if I want it to complete the line. It is software built for writing code, but it "understands" English well enough that I could occasionally just hit TAB and move on.

And thus, the real debate begins. If I hit TAB a single time while writing this post, did I use AI to generate the post?

The core problem with demanding blanket AI disclosures across creative disciplines, be it writing, game design, digital artwork, or video production is that it punishes the wrong people. The definition of "AI" is loose and poorly understood. To compound this, the word "use" is entirely subjective and varies wildly from person to person.

Are we dealing with an automated factory assembly line, or a power tool in a woodworking shop?

The Implication: "Zero AI Tolerance"

The most reasonable interpretation of these disclaimers is that creative gates (agents, publishers, etc.) want to know if the core output was generated by a machine. In prose, they want to know if an LLM wrote the text. In game development, if the code or sprites were generated wholesale. In video, if an avatar replace the creator or AI generated effects are used.

That is a reasonable stance. Wholesale automated output strips away the human element of what should be a very human venture.

The problematic aspect, however, is the "research, brainstorming, or ANY OTHER TASK" clause. "Any other task" is the broadest phrasing imaginable.

If an agency, studio, or publisher truly enforces a strict "no AI in any step of the process" rule, their only viable clients are living off-grid using typewriters, pencils, and having essentially zero digital tooling. If a someone searches for literally anything on Google, the first thing they will see and read is an AI synthesized search summary or answer to the query. If a solo game developer queries an engine error, an AI summary parses the documentation for them and answers the problem.

So, If at any point during your entire creative endeavour you searched a single reference item on major search engines (Bing included) while building your world, designing your character, or scripting your video, you used AI. You are now technically disqualified.

So what actually constitutes "use," and where do we draw the line between using technology and abandoning human craft?

Grammarly launched in 2009 as a basic grammar check. Today, it advertises itself as an AI-powered assistant. Digital art programs like Photoshop have spent years quietly integrating machine-learning algorithms for smart fills, edge detection, and color matching. Audio software uses "AI" and machine-learning algorithms to auto-level audio tracks or clean up background noise for video producers.

Use the smart-select tool in your art software? Use a denoiser on your podcast audio? Use basic spellcheck in Microsoft Word? Technically, you used AI. You are disqualified?

The Honesty Penalty

Many people are fundamentally unfamiliar with how modern software works. "AI" to them is a vague buzzword. If you say "large language model" or "diffusion," they look at you confused. This is true for many brilliant artists, writers, and musicians who are deeply creative but not inherently tech-savvy.

As someone who works closely with technology, I probably know more about how these underlying AI systems function than your average person. When a submission form asks, "How much AI did you use in your project?" I have to stop and really think.

I searched thousands of obscure reference details while writing my book, hitting AI-summarized web results every time. Did I use AI thousands of times? Yes..?

Someone who doesn't track tech developments or who doesn't fundamentally understand what "AI" is and isn't wouldn't think twice about their Google searches. They would simply check "NONE" and move on.

The dilemma comes down to interpreting intent:

If you detail every ambient instance of technology, you risk immediate, automated rejection by an evaluator who doesn't understand the difference between a smart search filter and an automated generator. If you say "None," you are technically lying by their broad definition.

The Blurry Line of Assistance: A Creative Ship of Theseus

Let's look at how human modification and machine iteration blend together. Consider a short narrative snippet, though this exact process applies to refining a game mechanic, tweaking a shader, or re-wording a voiceover script.

Here is a quick draft written entirely by hand in under a minute:

Bob is a builder. He lives in a house with his family. Bob wants to write a book. Every night, after his family is asleep, he sits at his desk. He types. He writes his story.

It is basic and flat. Let's run it through a generative tool for a rewrite.

Prompt: Rewrite this to be more atmospheric and literary.

By day, Bob shaped the physical world with wood and stone. He had constructed a good life, anchored by an elegant home that hummed with the warmth of his family. Yet, a different kind of architecture occupied his mind. Every night, after the quiet rhythm of sleep settled over the house, he would descend the stairs in silence. The office became his sanctuary, a pocket of darkness broken only by the cold glow of the monitor. He sat behind the desk, rested his hands on the keys, and built a new world.

That feels overwritten. Let's dial it back manually, stripping out the overly flowery prose and rewriting the blurb by hand:

By day, Bob was a builder. He worked with wood and stone, building a quiet life for his family. Yet a different type of work occupied his mind. Every night, after the house fell silent, he would slip downstairs to his office. Illuminated by the glow of his monitor, he would type away the midnight hours. Just as by day he constructed physical structures, by night he was constructing entire worlds.

Now, let's take that human-edited version and pass it through a tool one last time for minor polish.

Prompt: Keep this almost identical, just smooth out the transitions.

By day, Bob was a builder. He worked with wood and stone, crafting a quiet life for his family. Yet a different type of work occupied his mind. Every night, after the house fell silent, he would slip downstairs to his office. Illuminated by the pale glow of his monitor, he typed away the midnight hours. Just as by day he constructed physical structures, by night he constructed entire worlds.

And a final human touch:

By day, Bob was a builder. He worked with wood and stone, crafting a quiet life for his growing family. Yet a different type of work occupied his mind. Every night, after the house fell silent, he would slip downstairs to his office. Illuminated by the pale glow of his computer monitor, he typed away the midnight hours. Just as by day he constructed physical structures, by night he constructed entire worlds.

We now have four versions of the same concept.

  1. The first is 100% human.
  2. The second is largely machine-generated, derived from a human input prompt ("seed").
  3. The third is human-restructured prose built from the machine's draft.
  4. The fourth is a slight technological polish of the human edit.
  5. A fifth edit with tiny tweaks by human

Which parts are human? Which parts are machine? How many times can this iterative process go on for until the lines between human and machine are erased?

It becomes a creative Ship of Theseus. Every plank has been swapped, adjusted, or sanded down. This same blur can happen across many different mediums as well. If you use AI to clean up your code for a video game, does that make your game worthless AI slop? How much "AI" is too much AI. If you wrote a beautiful code function for 45 minutes, then have AI clean it up and evaluate it for performance gains, is it no longer your code? Did you just shoot yourself in the foot even if the performance of your game just increased by 10%?

With writing, automated "AI detectors" claim to spot these differences and flag AI generated text, but they are notoriously unreliable. They frequently flag 100% human-written classical literature as machine-generated, operating primarily as snake oil designed to sell "humanizing" subscription services. For example, one website I tested flagged the Bob story text as 92% AI...but we just showed that to be false, I wrote well over 50% of it, even if we're being conservative. To further prove my point, I submitted 2 paragraphs from chapter 1 of Jurassic Park (the novel) and it returned ... 82% AI. Like I said, very accurate. Michael Crichton was ahead of his time with LLM use in 1990.

Chisels vs. Planers: Augmentation as Craft

Prompting, refining, and integrating technology effectively is a skill. It requires taste, discernment, and an understanding of the underlying domain.

Today, even with cutting edge models like Fable 5, you cannot simply ask an AI: "Write me a best-selling novel," "Build me a AAA game," or "Paint me a gallery-worthy piece." The output of a single raw prompt is almost universally flat, derivative, and recognizable as "slop".

However, using technology to brainstorm edge cases, test color palettes, debug code, or refine awkward phrasing is an entirely different workflow. Used correctly, it is an augmentation of human skill, not a replacement for imagination.

Nobody criticizes a woodworker for using a electric planer instead of a hand chisel, or a table saw instead of a handsaw. The power tool speeds up the tedious removal of raw material, but the design, the joinery, the assembly, and the final finish still belong entirely to the craftsperson.

Yet in many creative fields, there is an expectation that artists must continue using hand chisels for every step of the process, or risk losing their legitimacy. AI is now, in many ways, a swiss-army knife power tool for creative ventures.

Creative Panic and Institutional Fear

It is easy to understand where the panic comes from. Creative industries view modern AI development as an existential threat. If automated tools can generate passable assets, what happens to entry-level concept artists, junior coders, copy editors, layout designers, and voice actors?

The fear is real, valid, and justified. Unchecked automation in the hands of cost-cutting executives threatens to flood platforms with low-effort content.

This creates a sharp divide in how different industries view these tools. In tech and software engineering, finding a tool that speeds up production or cleans up code is praised as an efficiency gain. In traditional creative industries, using the exact same class of AI tooling is often treated as a moral failing or a legal liability.

This stance stems heavily from two factors:

  1. Legal uncertainty: Copyright law regarding synthetic output remains muddy. Studios and publishers fear signing work that cannot be legally protected.
  2. Reputational risk: Gatekeepers fear accidentally greenlighting low-quality automated work and facing backlash from audiences who value human provenance.

Trying to ban modern computational tools from creative workflows is like trying to ban autotune from modern music production or digital drawing tablets from illustration. It is an impossible holdout.

Blanket bans also act as a form of gatekeeping. An independent game developer who cannot afford $5,000 for custom sound design might use AI audio tools to prototype soundscapes. An aspiring author who cannot afford thousands of dollars for developmental editing from a "pro" might use a language model to point out pacing lulls in a chapter. A solo creator working in a second language might use tools to ensure their grammar translates clearly to a more marketable language.

If the final output is meaningful, cohesive, carefully crafted, and directed by a human mind, does the specific brand of power tool ruin the art?

The Unavoidable Path Forward

As creators navigate submissions, client work, and publishing, they face an awkward reality. Institutional rules are often written by people who do not fully understand the technical mechanics of the software their creators use every day.

Clear, mature guidelines are needed. A constructive partner does not ask a vague, binary question like "Did you use AI?"

Instead, industry standards need to define acceptable boundaries:

Pandora’s box is open. We can't close it now. We just have to adapt. Younger creators growing up alongside these tools will not view them as disruptions, but as standard features of their tooling.

The question should never be whether a tool was present in the studio. The question should always be: How did you use your tools? Who directed the work, who supplied the vision, and is the final creation worth existing?

As for my own work? No, an LLM didn't write my book, design my projects, or build my ideas. I did.

But I certainly used every modern tool available to learn, research, and sharpen the final product. And I'm not going to pretend otherwise because that would make a liar.