As AI Floods the Video Feed, Be Prepared for People to Watch Less

Video is winning the internet. That part is settled. What is not settled is whether audiences will keep showing up for it at the rate the industry has come to expect, because the data from the past eighteen months tells a more complicated story. People are tired. They are deleting apps, posting less, skipping videos they say they want to watch, and quietly opting out of platforms that once felt unmissable. And this is happening at the exact moment that producing video content has become almost free.

I have a colleague who has trained ElevenLabs on a business owner’s voice and HeyGen on his likeness. The owner never sits in front of a camera. He never records a script. Yet daily videos go out under his face and in his voice, fully AI-generated (GenAI), indistinguishable to a casual viewer from the real thing. This is not a science project. It is a Tuesday.

Disclosure: It’s working fantastically for this business… for now.

As the cost of producing convincing video approaches zero, the supply curve has gone vertical, and every feed on every platform is being filled faster than humans can keep pace.

The Fatigue Is Measurable Now

The interesting question is no longer whether AI can generate the content. It is whether anyone will want to watch it. Deloitte’s 2026 Digital Consumer Trends report is the clearest baseline we have:

70% of people wish they spent less time on their devices. 1 in 4 have deleted a social media app to take a break. Gen Z and Millennials show the highest concern for digital fatigue. Usage intensity is striking: 34% check their phones at least 50 times daily, while 15% do so over 100 times.

Deloitte Digital Consumer Trends 2026

A separate analysis cited across digital wellness coverage found that adult digital burnout is 62%, broken into three loads the human nervous system was never designed to carry simultaneously: information overload, communication overload, and social comparison overload.

The behavioral shift on the production side is just as telling. The BBC found that nearly a third of social media users post less than they did a year ago, and OneCom Media translated that finding into something marketers should sit with:

Many users are sharing less and spending more time watching than posting. People are still online, but a growing number are opting out of the pressure to “perform” publicly. They are consuming content, not contributing to it as much, and that changes what audiences respond to.

OneCom Media & Marketing

1. Surface engagement drops while real intent remains present underneath, now expressed through saves, DMs, and silent comparisons rather than likes and comments. This breaks the dashboards that most brands still use to judge whether their content is working.

The Detox Studies Are Getting Harder to Dismiss

The November 2025 study published in JAMA Network Open is worth knowing about by name, because it used passive phone monitoring rather than self-reporting. Researchers at Beth Israel Deaconess tracked 295 adults aged 18 to 24 through a one-week social media break and watched what actually happened on their devices. The mental health gains were significant. Depression dropped 25%, anxiety fell 16%, and insomnia decreased 14% during the detox week. NPR’s coverage of a parallel three-week intervention found similar numbers:

At baseline, the participants were spending about two hours a day on the five social media apps the study was looking at. During the detox, it fell to like 30 minutes a day. By the end of that third week, the team found a 16% reduction in anxiety symptoms, a 24% decrease in depression symptoms and a 14.5% decrease in insomnia symptoms.

NPR

There is a larger study in the field right now, led by Steven Rathje at Carnegie Mellon, with over 8,000 participants across 23 countries, each limiting use of TikTok, Instagram, X, and Facebook to 5 minutes per app per day for 2 weeks. Findings are expected early next year. If the smaller studies are predictive, the results will not be flattering to the platforms.

The Real Story Is Choice, Not Attention

There is a tired narrative that human attention spans are shrinking, usually accompanied by a goldfish comparison that has been debunked more times than I can count. People still sit through three-hour movies. They still read long novels. They still listen to four-hour podcasts on subjects they care about. The capacity has not changed.

What has changed is the size of the menu. A viewer in 1995 had a few dozen meaningful entertainment options at any given moment. A viewer in 2026 has effectively infinite ones, algorithmically queued, refreshing every few seconds, each one engineered to be marginally more compelling than whatever they were doing before. When 52% of viewers report skipping videos longer than 60 seconds, even when the topic interests them, that is not a cognitive limitation. That is a rational response to abundance. The opportunity cost of staying on any single piece of content has gone up dramatically because the next thing is always one swipe away, and the algorithm has already decided it might be better.

This is the part that should make video marketers uncomfortable. The average Facebook session ran 2.7 minutes in 2013. By 2025, it was 54 seconds. People are not less capable of paying attention. They are more capable of leaving. And as choice continues to explode, the bar for what earns sustained viewing keeps moving up.

The AI Flood Is Pouring Into a Bucket That Is Already Full

Here is the collision course. The cost of producing video has fallen by about two orders of magnitude in the past 18 months. ElevenLabs clones a voice from a few minutes of audio. HeyGen produces a passable talking-head avatar from a short reference clip. AI can spin up B-roll that would have required a film crew two years ago. The marginal cost of one more video, in many workflows, is now closer to the cost of an API call than the cost of a production day.

Which means the choice explosion is about to become a choice supernova. Whatever the volume of content competing for attention was last year, it will be many multiples of that by next year, almost all of it generated rather than created. Deloitte saw this coming in their 2026 Media and Entertainment Outlook, where they were direct about the consumer reaction:

AI fatigue sets in as consumers — especially younger audiences — moderate daily engagement on social platforms.

Deloitte 2026 Media & Entertainment Industry Outlook

So the supply of video content is exploding, the audience is fatigued and increasingly selective, and the platforms themselves are starting to register a measurable drop in engagement among the very demographics that defined them. The math here does not work for anyone whose strategy is to publish more.

The Honest Counterpoint

Two things deserve to be said out loud, because they complicate the easy narrative.

The first is that detoxes do not necessarily reduce screen time. They redistribute it. The JAMA study found that participants who quit the major social apps cut about nine hours per week from those platforms but increased overall phone time by roughly fifteen seconds per day. Streaming, games, and texting absorbed the displaced attention. The phone is an addiction. The apps are interchangeable.

The second is that the digital detox has itself become content. Lifebonder put it more bluntly than most:

Irony alert: the majority of individuals detox from social media… and then immediately post about it. They vlog it, make detox challenge videos and they get engagement from not engaging. Social media detoxing, ironically, becomes content. The machine adapts. Your resistance is part of the marketing cycle.

Lifebonder

Both of these are true. Both also miss the deeper point. The aggregate signal from people who are not posting about their detox, who are simply quietly closing the app and going outside, does not show up in any feed.

A Personal Note

I am one of those people. I watch less than I used to. I post less than I used to. My wife and I walk twice a day with the dogs, and I am averaging ten thousand steps without trying. I am down thirty pounds from where I was a year ago. I spend evenings in the yard landscaping rather than in a feed, and I cannot remember the last time I felt like I missed anything by being absent.

Nobody optimized me into this. No app suggested it. There was no challenge, no streak, no creator who convinced me. I gradually noticed that the time I was spending watching other people live was time I wasn’t spending living, and the trade stopped feeling worth it. I suspect many people are arriving at this conclusion through their own private routes, without writing a thread about it.

So, Is the Video Explosion Going to Be Worth It

For most brands, no. They’re missing the bump.

The economics of AI-generated video assume that the constraint is production cost. But the constraint was never production cost. The constraint is human attention, and that supply is genuinely fixed. Pouring more content into a feed where viewers are skipping most of what they actually want to watch does not increase reach. It produces more noise, faster fatigue, and a quicker path to the moment your audience closes the app.

The brands that will do well in this environment are the ones that notice that observer mode is not disengagement, it is filtration. People are still watching. They are just being far more selective about whose voice and whose face earn those few seconds of consideration. Authenticity, founder presence, employee voices, and customer stories are no longer soft virtues. They are the only signals strong enough to break through a feed that is otherwise filling up with synthetic versions of everyone else.

Marketing has always been the discipline of differentiation. The job is to find the angle, the voice, the perspective, or the experience that nobody else can credibly offer, and to use it to earn attention in a market that grants none for free. AI-generated video is the exact opposite of that. It is the entire industry pointing the same tools at the same templates and producing the same talking-head clips with the same synthetic delivery, then wondering why nothing breaks through. There is no differentiation in adopting the tool that everyone else has just adopted. There is only faster average convergence.

The brands that will get noticed in 2026 are the ones that recognize that when everyone is doing the same thing, the unique opportunity is to do something else. Show up as a real person. Take a position that requires conviction. Please create a piece of content that an AI cannot produce because it depends on lived experience, original reporting, a relationship, or a point of view. The value of human-made marketing is going up precisely because the supply of synthetic marketing is going to infinity.

The rest of us will be outside.