I tried to optimize for AI search, but my Kansai dialect was the culprit | AI Utilization | Generative AI | Writing | #325

The trigger was this article

The other day, I read this article by Takamitsu Hamaguchi.

It's an article where a pharmacist breaks down and
explains arXiv papers, and the
content was incredibly interesting.

So, I also had an AI read and understand the original paper
(a joint study by the University of Tokyo, University of Tsukuba, and Hiroshima University).

"Couldn't I use this for KITAcore's note?"

...that thought was how it all began.

1. The paper's claim

"By optimizing just these three things, the citation rate in AI search increased by an average of +17.3%"

is the study.

The specific target values are here:

2. Does this apply to me too!?

I tried grading myself honestly.

Headings (Macro): Pretty much OK

KITAcore articles use H2 and H3 properly. The structure isn't bad.

Bullet points/Tables (Meso): OK for experimental articles

Data analysis and tool introduction articles
naturally include bullet points and tables.

Bold text (Micro): This is also reasonably OK

I use bold text for important points.

"Wait, am I actually doing okay?"

3. A fatal problem

It is written like this in the paper.

Clear and concise writing style.

My writing style:

………….

That's all it is.

The problem is just my dialect.

Even though the structure is relatively okay,
my writing style might be holding me back.

4. KITAcore from the perspective of AI search

Content that AI search engines want to cite:

✅ Stuff they want to cite

「構造化要素の比率は25〜35%が最適とされる」
「段落長が300語を超えると注意力が31%低下する」

❌ Stuff they don't want to cite

「これめっちゃ効くやんけ」
「ワイもアホやったわ〜」
「知らんけど」

AI: "Does this source seem authoritative?"

My article: "The time I was a total idiot"

AI: "………… (skipping citation)"

5. Maybe that's fine

Time to stop caring.

What I want to do on note is:

• Resonating with people who have the same problem

• Making them think, 'I'm not the only one'

• Making them laugh a little after reading

The moment I rewrite everything into standard Japanese and start writing things like 'According to the findings of this paper...'
it stops being KITAcore.

The fact that my Kansai-style 'wai-go' is hard to cite also means
there is less competitionas well.

It's fine if it's not cited, as long as it's read.

It's fine if it's not optimized for AI search,
as long as it reaches the people it resonates with.

That is the KITAcore way.

Probably.

I don't know, though.

6. Summary

The structure can be improved.

But I won't stop using my Kansai dialect.

This is the KITAcore declaration.

7. Bonus: Who can use GEO-SFE?

Since you're probably wondering 'So who can use it?', here it is:

• People writing how-to and explanatory articles

• People writing technical articles and documentation

• People managing internal documents with NotebookLM (Hamaguchi-san's article has details)

I will use it for HTML manuals.

I (almost) never use it on note.

That is all.

8. Summary of the paper

If you read this article and think,
"I was the same way" or "That was interesting"

please leave a like or a comment.
It won't be cited in AI searches, but it will boost my motivation.

It won't be cited in AI searches, but it will boost my motivation.

▼ Please be sure to read Mr. Hamaguchi's original article as well.

See you later.

▼ Click the image to check the weekly ranking.

▼ KITAcore's note status is now public: click the image to see the details.

#GenerativeAI #AIUtilization #AISearch #SEO #GEO #ChatGPT #ArticleDesign #WritingTips #InformationDissemination #PaperExplanation #note #Diary #Work #Business #FailureStory #Essay #Learning #Output #Creation #Self