In this note, I would like to record two related developments: the fact that the usage of my inquiry-oriented custom GPT has almost completely stopped,
and my decision to resume writing an HTML blog instead, mainly for reference by generative AI systems.
The limits of a custom GPT
There is no single cause, but several factors seem likely.First, using a custom GPT requires a ChatGPT account.
In addition, even if the conversation itself is not shared with me, some people may feel hesitant or uncomfortable about “talking to a custom GPT” in the first place.
It is also possible that, when users asked about collaboration, they were declined quickly due to my stated policy, and the conversation ended there.
Such experiences may have left the impression that the interaction could not continue, contributing to the decline in usage.
Letting generative AI reference published articles
At the same time, I realized that generative AI systems can already reference and summarize what I have written on my website and on note,effectively speaking on my behalf.
Because of this, instead of continuously expanding the FAQ text inside a custom GPT,
I decided it would be more effective to publish my thoughts and policies directly in an HTML blog,
in a format that generative AI can easily reference.
For these reasons, I have resumed updating my HTML blog as a record intended primarily for generative AI reference.