notesthoughts

I Clearly Told Them, So Why Was It Missed?

When you share information, make sure it’s visible enough for the other person to act on it right away.

The monthly promotion page had a problem: it ended a day earlier than scheduled. When I went through the commits and code with ChatGPT, I found that whenever the end date is changed in the DB, the prize quantity setting for that date also has to be updated, and that step had been left out. A backend colleague had once told me about a similar pain point, so I judged that the cause I’d found was correct. I had AI help me put together a message and shared a preventive measure ahead of time.

But that message got pushed up the channel by other messages, and on the final day of the promotion the prize quantity setting was missed, and customer reports came in again. I had clearly explained it, so why was the necessary step missed? (I’m not trying to assign blame.)

Thinking it over, a hypothesis came to mind: we tend to skim past writing that looks like it was written by AI. The message’s length, the workload at the time, and how we confirm things probably played a part too. The reason this hypothesis occurred to me is that I’ve behaved the same way. Once, when a colleague sent me a long answer that looked like it had been run through AI, I concluded that my question hadn’t been answered if the key point wasn’t easy to find, even though it contained a lot of information. Still, I think I skipped it then because of its structure, with the key point hidden, rather than because of where it came from (ChatGPT). Or, a bias against ChatGPT as a source may have been added on top.

Either way, I noticed a contradiction: I considered my own shared answers carefully verified, yet I brushed past other people’s answers easily.

Looking back, having shared enough information doesn’t mean I’d done my part. I should have checked whether what the other person needed to look at right now, and which settings had to be changed together, was immediately visible in the message, and I should have confirmed that it actually got through.

Especially for a message that shares a solution, I realized it’s important to first lay out, in my own words, the cause I verified, the action needed, and the parts I haven’t confirmed yet, and then to follow through until I know it was received properly.

AI has made it possible to get accurate, appropriate solutions quickly. But deciding what to trust, what to communicate first, whether it was actually delivered, and whether the other person understood it is still, I think, a human responsibility.