What people are saying
Nobody has said anything yet.
Brand New Thoughts/Creative Technology
I Built It With AI. That Doesn't Mean AI Built the Idea.
Generative tools can write code, generate images and suggest fixes. They still need somebody to decide what problem is worth solving and what a good result looks like.
By ElleJanelle|August 4, 2026|3 min read
"AI Made It" Is Usually Too Small a Description
During the live build that became the Community Scroll, ElleJanelle was bouncing between OBS, a website, image generation, HTML, Claude, ChatGPT and whatever tool could solve the next problem fastest.
From the outside, that can be summarized in one sentence:
She made it with AI.
Technically true.
Also incomplete.
The tool did not wake up and decide livestream audiences needed an infinite visual scroll where community members could upload photos, generate branded characters and become part of the broadcast.
That problem came first.
AI can help build an answer. Somebody still has to decide what question is worth answering.
Brand New Thoughts
The Work Moved Up a Level
Traditional production often requires direct manipulation.
You draw the element.
Write the function.
Mask the image.
Animate the transition.
Generative tools can move some of that work into instruction, selection and iteration.
Now the creator may spend more time saying:
No, not that.
Make it transparent.
The whole scroll needs to move.
The image is getting cropped.
The brand mark is too big.
The art style is generic.
The output technically works but does not feel like the show.
That is still work.
It is a different layer of work.
Taste Becomes More Valuable When Output Gets Cheap
If ten people can generate ten images in ten seconds, the scarce resource is no longer simply the ability to produce an image.
It becomes:
- knowing what to ask for;
- recognizing which result is wrong;
- identifying why it is wrong;
- preserving visual continuity;
- understanding the audience experience;
- combining several tools into one system;
- deciding what should remain human-controlled.
That is why AI can increase the value of taste at the same time it lowers the technical barrier to production.
What the Copyright Office says
The legal question is not identical to the creative question, but the U.S. Copyright Office has been forced to draw a related line.
Its 2025 report on generative AI concluded that copyright protection can apply to AI-assisted works where a human author contributes sufficient expressive elements, including creative selection, arrangement or modification, while mere prompting by itself is not enough. The Copyright Office's AI reports are here.
That legal standard is narrower than the broader question of who had the idea.
But it illustrates the distinction between requesting output and shaping a work.
AI Did Not Remove the Failure
One of the funniest parts of building with AI is how quickly the fantasy of effortless automation dies.
The code can be wrong.
The model misunderstands the instruction.
The image invents a face.
The database points at the wrong project.
The local server works while the OBS browser source does not.
The "fix" breaks something that was already working.
That is not an argument against AI.
It is an argument against treating AI use as the absence of skill.
Skill increasingly includes debugging the relationship between several systems you did not write from scratch.
The job title for this kind of work
ElleJanelle repeatedly describes herself as a creative technologist.
The phrase fits because the work does not sit neatly inside one traditional craft.
It is part producer.
Part designer.
Part developer.
Part art director.
Part product manager.
Part performer.
The goal is not to prove mastery of every tool individually.
It is to make the tools cooperate long enough to produce an experience that did not exist before.
The Better Question Than "Did AI Make This?"
Ask:
Who identified the problem?
Who chose the constraints?
Who rejected the bad outputs?
Who decided the final form?
Who connected the systems?
Who understands why the result exists?
The answers will vary from project to project.
That is more informative than a binary label.
Because "AI-generated" can describe everything from typing one sentence and accepting the first output to spending three days directing, editing and integrating hundreds of machine-generated components.
Those are not the same creative process.
Where does authorship live in AI-assisted work?
Add to the DiscussionWhat part of a creative project do you think matters most for authorship: the original idea, the execution, the selection, the editing, or the final arrangement?