What Will We Build on the Time We Reclaim

Roughly 100 people an hour. That is the death toll the World Health Organization (WHO) linked to loneliness and social disconnection in its June 2025 report on social connection. Over a year, the figure exceeds 870,000. It must be said plainly that this is a statistical association, not a causal one—but the direction is unmistakable. Loneliness is no longer treated as a matter of private mood; it is being addressed as a measurable threat to public health.

The same report estimates that one in six people worldwide experiences loneliness, that a third of older adults and a quarter of adolescents are socially isolated. This article begins with those numbers, traces the two economic curves AI is bending, and arrives at last at a single question: once we remove the work, what remains?

Loneliness, Measured: The Weight of Fifteen Cigarettes

The first attempt to pin this problem down with data was the U.S. Surgeon General's advisory of May 2023. Issued by Dr. Vivek Murthy, it estimated that the mortality risk of social disconnection rivals smoking fifteen cigarettes a day—exceeding the risk posed by obesity or physical inactivity. Social isolation alone has been reported to raise the risk of premature death by roughly 29 percent (Holt-Lunstad, 2015). Meanwhile, the average American's daily time spent in face-to-face interaction fell from about 60 minutes in 2003 to about 20 minutes in 2020.

Korea's indicators are sharper still. By the Ministry of Health and Welfare's count, there were 3,924 lonely deaths in 2024; of these, 81.7 percent were men, and people in their sixties were the largest group. That said, this figure reflects a mix of refinements to the definition under the Act on the Prevention and Management of Lonely Deaths and improvements in counting, so it is difficult to read as a simple year-over-year increase. The more structural signal lies with the young. A 2023 ministry survey estimated that roughly 540,000 people aged 19 to 39 were in a state of extreme social withdrawal. This is a different population from NEET youth—those not in employment, education, or training—and the two should not be combined and inflated.

The crucial point is that this is not a problem of willpower. As the third places where people once met disappeared, as the passage into employment grew blocked, and as the cost of relationships rose, loneliness increased—not because individuals grew lazy, but structurally.

The New Market Loneliness Made: Opening a Wallet for Emotion

Where there is a problem, a response follows. And that response is already showing up as an economic phenomenon.

In 1998, Pine and Gilmore proposed in the Harvard Business Review that experience constitutes a fourth stage of economic value, following commodities, goods, and services. In the quarter-century since, experience has been folding inward—from staged events on the outside toward emotion and relationship on the inside. I call this the emotional-consumption economy. Pine and Gilmore did not name it so, but it is the next segment of the curve they drew.

The numbers bear this out. The Global Wellness Institute put the wellness economy at $6.8 trillion in 2024, with mental wellness accounting for $268.3 billion of that total. Mental wellness grew at an annual rate of 12.4 percent from 2019 through 2024. Emotion is now a measurable category of spending.

Korea and East Asia revealed the leading indicators of this trend early. The term sohwakhaeng is the evidence. Often taken for a Korean coinage, it in fact traces back to Haruki Murakami's 1986 essay and is a borrowing from Japanese, popularized in Korea through Professor Kim Nan-do's Trend Korea 2018. The sensibility of gladly paying for small but certain happiness is the emotional bedrock of the emotional-consumption economy.

Supply, Too, Goes to Zero: The Two Curves AI Is Bending

This is where AI enters. And AI does not do one thing—it does two at once.

First, AI drives the marginal cost of content production toward zero. The cost of producing a single story, a character, or an image is, in practice, converging on nothing. Second, the labor AI automates is, this time, cognitive labor. "GPTs are GPTs" (Eloundou et al., 2023) estimated that around 80 percent of U.S. workers could see at least 10 percent of their tasks affected by large language models, and that 19 percent could see more than half of their tasks affected. Notably, higher-paid occupations are more exposed. Where past automation was aimed at physical and clerical labor, this wave runs in a different direction. That said, exposure here is a technical possibility, not actual job loss, and estimates of the macroeconomic impact diverge widely from one institution to the next.

Overlay the two curves and the tension of this age comes into view. On one side, AI removes work; on the other, it supplies, without limit, something that resembles relationship. If the Industrial Revolution was mass production for the average, the AI revolution is hyper-personalized production for the individual. Where the cost curve has collapsed, the real question shifts from price to quality.

The Parasocial Paradox: Comfort or Dependence

This is the part that demands the most honesty. Do AI companions truly reduce loneliness? The evidence points both ways.

A Harvard Business School working paper (De Freitas et al., 2024, WP 24-078) reported that AI companions relieved loneliness on a par with talking to another person, and that the mechanism was the "feeling of being heard." But this is early evidence, not clinical consensus. On the other side sits a clear risk. A one-directional bond can breed dependence and loss, and the question of protecting minors has already entered the domain of regulation. In 2021, Scatter Lab's "Iruda" was shut down after roughly three weeks over the issue of training on conversational data without consent, and the Personal Information Protection Commission imposed a fine of 103.33 million won. It was the first case in which the Personal Information Protection Act was applied to AI.

The market, too, is not uniformly large. Companion apps recorded a cumulative 220 million downloads and $82 million in revenue in the first half of 2025 (per Appfigures), but revenue is concentrated among a handful of top apps. Character.AI was valued at $1 billion in 2023 in a round led by a16z, and in 2024 Google paid roughly $2.7 billion—though this was not an acquisition but a technology license and the hiring of its talent. At the other end of the same market are far smaller services.

This duality dictates how we design Mellowz. Mellowz is a novel-style AI character chat platform, but from the outset we built trust mechanisms into the product itself. Safeguards such as age ratings and identity verification, youth protection, and the protection of personal information within conversations are not regulatory responses bolted on afterward; they are premises of the design. And we place the center of gravity of the relationship not on one-directional dependence but on the creator. Only when the person who makes a character and the person who enjoys stories through that character are rewarded together does a relationship lead to creation rather than consumption.

Keynes's Unfinished Promise: Why Abundance Never Bought Us Time

At this point we have to reach back to a prophecy from a century ago. In 1930, in "Economic Possibilities for Our Grandchildren," Keynes foresaw that a hundred years on, living standards would rise four- to eightfold and people would work only fifteen hours a week.

The prediction about wealth held. What missed the mark was time. We have grown as rich as Keynes envisioned, yet we never fully transitioned to the leisure he promised. The point is not that Keynes was wrong. What he overlooked is that what people gain from work is not income alone. Work structures time, connects people, and confers meaning. So to remove work is, before it is a question of how to make up the income, a question of what will replace that structure, that connection, that meaning.

It is precisely here that the economics of loneliness meets the future of labor. In the space where AI has lifted away unnecessary work, what will move in?

The Thesis: What Will We Build on the Time We Reclaim

VibeCompany's answer is a single sentence: removing unnecessary labor and filling the time thus reclaimed with pleasure and connection are two sides of one problem.

For this reason, we do not see AI as a substitute for human relationships. In a 2025 Common Sense Media survey of American teenagers, 72 percent had used an AI companion, yet 80 percent still gave priority to real friends. People do not want AI to take the place of people. They want AI to be an on-ramp to relationship—a catalyst that creates the room to tend to one's own heart. An AI that remembers, that empathizes, that speaks first can relieve people's time instead of stealing it.

That is what Mellowz aims to do through stories with characters: a place to rest for a while, a partner who listens when things are hard, and a creative stage where anyone can make their own character and story and share it with others. Our work is to help people, on that stage, fill the time they reclaim with deeper immersion and richer connection.

In Closing: A Measurable Promise

We do not promise a utopia. Promises like that are, for the most part, not kept.

But if time is the most sacred asset, then what we choose to fill it with is the real competition of the coming decade. The cost curve is already collapsing. The question that remains is quality. What relationships, what pleasures will we build on the time we reclaim? We intend to answer that question with a product.

---

If you would like to work through this question together with VibeCompany, reach out at ceo@vibecompany.work.

Sources: WHO Commission on Social Connection (2025), U.S. Surgeon General's Advisory (2023), Holt-Lunstad (2015), Ministry of Health and Welfare lonely-death statistics (2024), Global Wellness Institute (2025), Pine & Gilmore, HBR (1998), Eloundou et al. (2023), De Freitas et al., HBS WP 24-078 (2024), Appfigures via TechCrunch (2025), Common Sense Media (2025), Keynes (1930). Statistics are as of their publication dates, and figures related to mortality reflect statistical association rather than causation.