Leaving out an image to keep the story moving
While preparing the dootang chapter about sharing proportions and fairness, I wanted to explain that the numbers follow an agreement people choose. They do not decide whether that agreement is fair. A calculation screen was one possible way to make this concrete.
The available image still contained sample names, so it was not ready to use as it stood. Having an illustration and having one ready to use were different things.
The attempt did not meet the requirements
AI helped prepare a version with the names covered and check the result. The preparation record noted that a name remained and the image dimensions had changed. That version was therefore not selected as evidence.
This does not establish that AI can never redact information. It establishes that this particular attempt did not meet the requirements for covering the intended information and preserving the source.
Choosing to continue with text
Returning to what I wanted to explain, the options were not limited to making the image work or stopping the whole chapter. The source of the proportions and the limits of the numbers could also be described in text, without using an image that had not met the requirements.
AI prepared the proposal and bilingual copy; I accepted the content and the text-led composition. The trade-off was that readers could not inspect the sample calculation screen for themselves in that chapter. The chapter could nevertheless continue without using an image that was not ready.
What I want to return to
This version of the dootang chapter therefore tells the agreement story in text and has been reviewed and accepted within its delivery scope. Acceptance does not establish that readers understand it better. What I want to keep from this event is the reason for stopping the image attempt at that point: an image should serve the story, rather than become a condition that prevents it from continuing.