I was practising reading code. Yet what I kept typing was Japanese.
This was during a SQL lesson with Kimiko-chan—my nickname for ChatGPT, not a human teacher.
I have spent years working with the programming language RPG. Now I was practising how to read SQL.
SQL is a language for retrieving and manipulating information in a database. At work, I had been dealing with a SQL script more than 1,000 lines long.
It was frightening—so intimidating that, to me, it felt like a haunted house. Whenever I mention a “haunted house” in this story, that long SQL script is what I mean.
Having experience and being able to read everything effortlessly are two different things.
In this session, I was describing conditions involving employee and department information in Japanese.
An employee’s department code had no matching code in the department list. In another explanation, I described finding departments in a particular region and retrieving the employees who belonged to them.
I was putting into words what was being compared and which conditions determined what to select.
Simply replacing each SQL term with a Japanese word would not produce a coherent explanation. I needed to follow the connections: which information was being examined, and what was being checked?
Something can seem clear while I am reading it. Explaining it myself makes me think about how the pieces fit together. In this exchange with the AI, it was my turn to do that explaining.
At one point, I wrote:
“What?! We’re still only at the entrance to the haunted house?!”
If that script of more than 1,000 lines was a haunted house, practising how to read these conditions was like taking my first steps through its entrance.
I thought I was making headway. Apparently, I was still only at the entrance.
My next response, though, was:
“All right! Bring it on!”
Far from running away, I was ready to go further.
Our collaboration that time was not about finishing a production program. The conversation with Kimiko-chan had become a place to learn. I read the conditions, described them, and kept going.
The AI being able to produce an explanation is not the same as me being able to explain it. When it is my turn to put things into words, my brain gets a share of the work.
There was, however, a problem with keeping that brain at work.
My eyelids.
At the end, I said I was enjoying myself and wanted to continue, but my eyes were starting to close. The way I put it was:
“My field of vision is getting pretty close to a single horizontal line.”
The enthusiasm was there. The available screen height was not.
So that was where I stopped for the session.
This is not a triumphant account of mastering SQL. I was trying to understand, explaining things in my own words, and wanting to continue. For me, that was an enjoyable lesson in itself.
I hadn’t fled the haunted house.
My eyelids had simply pulled down the shutters first.