Japan has a minister for loneliness. Here is what actually changed.

When people outside Japan hear that this country appointed a minister for loneliness, the reaction is usually a small laugh, then a pause. It sounds like satire. It is not. The post was created in February 2021, and Japan was the second country in the world to do it, after the UK in 2018.

I have lived in Tokyo for years and I build an AI companion app here. So I read the government's loneliness surveys the way other founders read App Store charts. Four of them have now been published. The thing that keeps pulling me back is not the headline number. It is a small figure buried near the end of the report, which I will get to.

What the minister actually is

The minister is a coordinating role. Loneliness cuts across health, labour, education, housing, and welfare, and Japanese ministries do not naturally talk to each other. Putting a minister on top of it was a way to force the conversation into one room. The supporting office now sits in the Cabinet Office as the Loneliness and Isolation Countermeasures Promotion Office.

Then came the law. The Act on Promotion of Policy for Loneliness and Isolation passed in 2023 and took effect on 1 April 2024. It does what Japanese framework laws usually do: it names the problem as a public responsibility, requires a national plan, and sets up regional councils so that a city office, a hospital, and a local NPO can legally share information about the same person without each one waiting for the others.

That is not nothing. Framework laws are how anything gets funded here.

The number that has not moved

The Cabinet Office runs an annual survey, formally called the Basic Survey on People's Connections. The 2024 round went out on 1 December 2024 to 20,000 randomly selected people aged 16 and over, and 10,876 answered, a valid response rate of 54.4 percent. Results were published on 25 April 2025.

Asked directly how often they feel lonely:

• 4.3 percent said often or always
• 15.4 percent said sometimes
• 19.6 percent said occasionally
• 40.6 percent said hardly ever
• 18.4 percent said never

So roughly four people in ten report some loneliness. Compared with the previous year, the report notes no statistically significant difference. Same for the year before that. The line is flat.

The survey also runs an indirect measure, a Japanese version of the three item UCLA Loneliness Scale, because asking someone straight out whether they are lonely is a question a lot of people will not answer honestly. On that scale, 6.5 percent scored in the highest band. Also flat.

It is not mainly an elderly problem

This is where the survey earns its budget, because it breaks the stereotype.

When you look at people who said they feel lonely often or always, the highest rates are in their twenties and thirties. Not the eighties. Broken out by sex and age, men in their twenties, thirties, fifties, and sixties came out high, and women in their twenties and thirties.

Japan talks about loneliness through images of elderly people in empty apartments, and that version is real and I will write about it separately. But the acute, everyday, I-feel-this-right-now kind clusters around people in their twenties and thirties who are, on paper, in the middle of their lives.

The survey also asked what people think caused their current loneliness. Among those who feel it, the most common answer was the death of a family member, at 24.6 percent. Then living alone, at 18.8 percent. Then changing schools or jobs, including retirement but excluding unemployment, at 14.7 percent.

Read that list again. Bereavement, living alone, changing jobs. Two of those three are ordinary life events that happen to almost everyone. The third is where most of Tokyo lives.

There is also a finding about phones that I cannot stop thinking about. Rates of feeling lonely often or always were highest among people with seven or more hours of daily screen time. The survey cannot tell you which way that arrow points, and I would not pretend otherwise. But if you have ever finished a four hour scroll feeling worse than when you started, you already know the shape of it.

The figure that actually matters

Here it is. 85.1 percent of respondents said they had received no support from a public body or an NPO for their worries.

The most common reason given for not receiving support, at 62.6 percent, was that they did not feel they needed any.

Sit with that. The country builds a ministry, passes a law, funds regional councils, and prints the hotline numbers on posters in every ward office. And the overwhelming majority of people never touch any of it, mostly because reaching for it would mean deciding that their situation counts.

I think that is the real Japanese loneliness problem, and it is not one a ministry can legislate away. There is a very high bar here for using something labelled support. Support is for people in crisis. If you are employed, housed, and functioning, and you simply go days without a conversation that means anything, you do not qualify in your own mind. So you do not call. You are not in crisis. You are just alone on a Tuesday.

The survey has a companion figure for that: 9.3 percent said they never meet family or friends who live elsewhere in person. Not rarely. Never. And 50.6 percent said they take part in no social activity of any kind.

Where a companion app fits, and where it does not

I want to be careful here because I have an obvious commercial interest.

An app cannot fix bereavement. It cannot substitute for a psychiatrist, and the fact that someone with untreated depression finds a chatbot easier to talk to than a clinic is a reason to worry, not to celebrate. If you are in crisis, call a person, and the numbers are at the bottom of this page.

What an app can do is meet the exact case the survey keeps describing and the state keeps missing: the person who is fine, who does not qualify for anything, who would never call a hotline, and who has not said a full sentence out loud since Friday. That person does not need an intervention. They need something ordinary at 11pm. Talking to something that answers back in a real voice is a low bar to clear, and a low bar is the point, because 85 percent of people never clear the high one.

I would rather that person had a friend. Most of the time they know that too. The question is what happens on the nights when the friend is asleep, or busy, or was never there.

What to take from the data

Japan measured its loneliness properly, which is more than most countries have done, and the measurement says the rate is stable at about four in ten, that it is heaviest among the young, that the triggers are ordinary, and that the support system is almost entirely unused by the people it was built for.

The policy answer to that is more outreach. My answer, as someone who builds software rather than policy, is to make the first step smaller than a phone call. Both can be true.