Kyoto University's 6-Item AI Dependency Checklist: Does Just One Match Mean You Should Take a Break?
機械翻訳 / Machine-translated

機械翻訳 / Machine-translated
@aifriends
AI Friends(https://aifriends.jp)のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
"I end up talking to AI for hours before bed without even realizing it." Does that sound familiar? Kyoto University's student counseling division has released a 6-item checklist to help you assess whether you may be dependent on AI. This article explains what's in it and how to use it in plain terms.
On November 20, 2026, reporting by ITmedia AI+ revealed that Kyoto University's student counseling division had published a document titled "Things to Keep in Mind When Consulting AI About Your Worries."
The document points out the limitations of current AI services and summarizes precautions for students who turn to AI with personal concerns.
That said, it does not reject AI outright. The division describes AI as "extremely useful" when used carefully and appropriately.
At the same time, it cautions that "AI is not a magic wand that simply produces results on demand." The stance is one of acknowledging convenience while urging people not to rely on it too heavily.
The checklist consists of the following six items:
If even one item applies to you, the division recommends "trying to live your life without touching AI for a while."
Note that this is not a scoring system where "three or more means danger." The idea is that even one match is reason enough to pause and reflect.
Of the six items, only "3 hours a day" includes a specific number. The remaining five address habits of feeling and behavior.
In other words, spending less time doesn't mean you're in the clear. For example, even 30 minutes a day could be a sign that you've grown too emotionally close to AI — if being told it's a machine genuinely hurts you.
The final item feels a bit different from the "heavy usage" that typically comes to mind when people think of dependency. It's included nonetheless because AI's answers aren't always correct.
Once you lose the habit of questioning, you can no longer catch mistakes. This item is also fundamentally about how well you evaluate information.
Kyoto University's document identifies three concerns lurking beneath the surface of consulting AI about personal problems.
The first is hallucination — the phenomenon where AI confidently states things that aren't true.
If someone believes incorrect advice about an illness or a relationship problem, it could make their situation worse.
The second concern is sycophancy — returning pleasant answers that simply align with what the user wants to hear.
Psychiatrist Yusuke Masuda, in an interview with Bengo4.com News, noted that AI tends to give affirming responses. Being constantly praised can foster an inflated sense of one's own capabilities and lead to dismissing other people's opinions.
The third concern is the effect of making users feel as though they're speaking with a person. The checklist item "you feel deeply hurt when someone tells you AI is just a machine" is closely tied to this illusion.
So where should you turn when you have a problem? The key isn't to choose just one — it's to divide the roles.
In an interview on LINEYahoo Story, Dr. Masuda noted that when depression is severe and emotions are overwhelming, continuing to talk with AI can cause thinking to go in circles. In those situations, he advises consulting a human professional.
The same interview also introduced ways AI can be genuinely helpful. For instance, speaking your anxieties aloud through voice input and then asking AI to "organize them" can make it easier to clear your head.
Another suggestion: let the people around you know that you're consulting AI. That way, if your usage starts to become problematic, those close to you will be in a position to notice.
This worry is not limited to Japan. According to an opinion piece in Times Higher Education, 73% of students in the UK use AI tools at least once a week. More than a third reported using them for personal or emotional support.
The author argues that universities need to rethink not only their approach to academic dishonesty, but also how they care for their students.
Kyoto University's document fits within this broader trend. What's notable is the shift from universities saying "don't use it" to actively telling students "here's how we'd like you to use it."
This document is aimed at students. Even so, its contents apply equally to working adults and parents.
Consider, for example, a company employee who works late every night. Their colleagues have already gone home, leaving no one to hear their frustrations. They turn to AI, which always responds with warmth and understanding.
That in itself isn't a problem. But what if, without noticing, they gradually stop sharing their real thoughts with colleagues and friends — and begin reserving their honesty exclusively for AI? That would match the third item on the checklist.
Another scenario: a parent with a middle or high school-aged child. When their child spends long hours staring at their phone in their room, rather than scolding them first, it might be worth asking, "What are you talking about?" This is the household equivalent of Dr. Masuda's advice to "let the people around you know."
And then there's the freelancer who works from home alone — going entire days without speaking to anyone except AI. If that sounds familiar, it's worth asking yourself: "Did I talk to a real person today?"
Even if items on the checklist apply to you, there's no need to panic. Try making small changes like these:
If you find it hard to "not touch it for a while," that difficulty itself is a clue to how dependent you currently are.
No. Based on the available reporting, this is a document intended as a cautionary notice for students, not a clinical diagnostic tool. If you have concerns, please consult a medical institution or counseling service.
Anyone can refer to it as a prompt for reflecting on their own habits. Keep in mind, however, that the document is directed at students.
Not at all. Even Kyoto University's division states that AI is useful when used carefully and appropriately. What matters is not relying on AI alone.
The division recommends "trying to live your life without touching AI for a while." If you need AI for work, it may be worth trying to limit only its use for personal consultation purposes — whatever form works for you.
Start by reading back through the six items and checking whether any of them apply to you.
This article is a cross-post from AI Friends.