The hours we are
not measuring.

There is a number I cannot stop turning over. The average American teenager spends eight hours and thirty-nine minutes per day on entertainment media. Not counting school. Not counting homework. Just the part of the day that used to be theirs.

I have spent twenty years building products that move billions of dollars and hours of human time. So I notice when an industry quietly absorbs a generation's afternoons and calls it growth.

We measure how many parameters our models have. We do not measure how many hours of childhood our products have absorbed. The first number we discuss endlessly. The second number is hidden in plain sight.

What the data say

Let me put the numbers down without softening them.

US teens, ages 13 to 18, average eight hours and thirty-nine minutes of entertainment screen time every day, according to Common Sense Media. That figure does not include time on screens for school or homework. It is purely leisure media — TikTok, YouTube, Instagram, gaming, streaming. Forty-one percent of US teens spend more than eight hours a day on screens. TikTok alone takes one hour and fifty-three minutes of the average teen's day — the largest single-app slice in the history of consumer software.

This is not a US-only story. The OECD's PISA 2022 study, surveying fifteen-year-olds across eighty-one countries, found that sixty-six percent of girls and sixty-one percent of boys spend three or more hours on social media on a typical weekday. Roughly one in four teens across OECD countries spend sixty hours per week or more on digital devices outside of school.

The variation across countries is striking. Average daily total screen time runs as high as ten hours and forty-eight minutes in South Africa and ten hours and twenty-seven minutes in the Philippines, and as low as four hours in Japan. The same global platforms reach all of these teenagers. The same recommendation algorithms run in their feeds. The same advertising auctions price their attention. And yet a teenager in Tokyo and a teenager in Manila are, on average, spending two and a half times more or less of their day on a screen.

I will come back to that variance, because it is the most hopeful number in this entire essay.

The story is not what they are doing. It is what they are not.

Here is what I cannot get past, and what I think we have collectively avoided saying out loud.

Those eight hours did not appear from nowhere. They came from somewhere. Time is the one resource we cannot manufacture more of. Every hour a teenager spends on a screen is an hour they are not spending on something else — and the something-else is the part of life that used to make them into the adult they were going to become.

Three numbers tell the story plainly.

Five hundred and nineteen minutes on screens. Thirty-nine minutes with a friend in person. Nine minutes reading. Whatever you believe about screens, those three numbers next to each other are the entire argument.

What got displaced

If the screen time is the new line, the displaced time is the line we should grieve.

The country gap is the most hopeful number in the dataset.

Now back to the variance. South Africa and the Philippines hover near eleven hours of daily screen time. Japan sits at four. That is a two-and-a-half-times spread between teenagers raised on the same global platforms.

This tells me something important. This is not destiny. It is the product of policy, culture, and infrastructure choices.

Japan's school system explicitly limits classroom phone use. South Korea has imposed cumulative gaming and social media restrictions on minors. France has banned phones in schools through age fifteen. The European Union's Digital Services Act imposes specific design constraints on platforms used by children. Each of these is debatable on its own merits. None of them are working perfectly. But together they prove a thing worth proving:

The trajectory of teen attention is not a force of nature. It is a product. Products can be redesigned.

What this means for the people building the products.

I spend my days building enterprise AI for the Global 10,000 — companies that ship AI for mission-critical, regulated workloads. The standards are not optional. We document model lineage. We monitor for drift. We govern data. We run continuous safety evaluations. Why? Because the cost of getting it wrong is regulatory action, financial loss, and broken trust.

Now look at what we build for children. Recommendation algorithms tuned to maximize watch time. Conversational agents filling the social gap teenagers used to fill with each other. Assessment tools that grade their work without context. Generative tools that finish their homework before they have finished thinking about it.

We are pouring AI into the formative years of an entire generation, and our standard of care is "the terms of service has been updated."

What I tell my own children.

I have three. Two are teenagers. One is not far behind. I do not pretend that I have figured this out. But here is what I have learned to say.

I do not start with "screens are bad." That conversation lost the room a decade ago, and it deserves to.

I start with: "What did your eight hours buy you today?"

If the answer is a friend they would not have otherwise spoken to, a piece of music that moved them, a video that taught them something they could not have learned in school, a window into a life that looks nothing like theirs — then those eight hours were investment. They were time spent, in the way an adult spends time on a thing they choose.

If the answer is a feed they cannot remember an hour later, a notification they answered without thinking, a string of strangers' lives they compared their own to and felt smaller for — then we have a problem we should both look at honestly.

And I am clear-eyed about the boring truth that some of these eight hours are mine to set the rules around. There is no algorithm that will design my family's evening. That is on me, and the same is true for every parent reading this.

What I am asking the rest of us for.

For the people building AI. The same standard of care you bring to a Fortune 500 customer should apply to a fourteen-year-old. If you would not ship a model unmonitored, ungoverned, and undocumented to a regulated bank, do not ship it that way to a child. The fact that the child cannot sue you is not a reason. It is the reason.

For the people regulating AI. Stop debating the existential. Start measuring the actual. The OECD has the framework. PISA has the instrument. Common Sense Media has the longitudinal data. There is no excuse to keep treating teen attention as a future problem when the current data are this clear.

For the rest of us. The question is not "screens or no screens." That is a fight from 2015. The question is which hours are real and which are filler — and whether the children we are raising will have any practice at the difference.

The most expensive thing we can spend is a teenager's hour. It does not come back. We should at least know how we are spending it.