The list of my achievements had grown, even though I hadn’t done any more work.
Wouldn’t that be convenient?
Unfortunately, that wasn’t what had happened.
I was working on a description of my past projects with “Kimiko-chan”—my nickname for ChatGPT, not a human co-author.
The material came from a system performance improvement I had carried out.
A process needed to retrieve matching records from a large body of data. It was taking too long, so I reviewed how it accessed that data and reduced the waiting time.
That was one piece of work.
But in the edited description, working with a large volume of data and making the process faster had become separate achievements.
One item for scale.
Another for speed.
They were two aspects of the same project.
I corrected the description, explaining that they belonged to the same project and should not be split into separate achievements.
What made this mistake tricky was that no completely unfamiliar project had appeared out of nowhere. I recognized the ingredients. The scale of the data and the improvement were both things I had described.
Yet changing the way those ingredients were grouped changed what the account conveyed.
Think of a dish. It is perfectly reasonable to describe its ingredients and cooking method separately. It is not reasonable to count them as two dishes.
That connection was what I wanted to restore.
What process had been causing trouble? What had I changed? What happened as a result? Those parts needed to stay together so the reader could understand the work.
There is one other distinction worth keeping clear.
I had made the original system improvement. My later collaboration with Kimiko-chan was about shaping the account of that experience.
It would be wrong to turn the original development into a project I had built with AI.
The AI organized the explanation. I checked how the pieces of the work fitted together and corrected the account where they did not. Making a description easy to read and making it faithful to experience overlap, but they are not the same task.
In this case, checking whether the numbers were right was not enough. I also needed to check which project each number belonged to, and what it described.
There was no need to multiply the work to make the achievement look impressive. One project could speak for itself—as one project.
Add headings if they help.
Just don’t let my work multiply along with them.