AI-Generated Short Content:The Next Big Threat to Our Attention Spans

What AI changes about the endless feed

A feed needs material, a way to choose between items, and an interface through which we watch them. Generative AI expands the first of those. Recommendation algorithms handle the second. A swipe makes the third almost effortless.

In September 2025, Meta announced Vibes, bringing AI video creation and remixing into a personalised short-video feed. It is a concrete example of these functions coming together, although the announcement tells us nothing about their effects on attention.

The ability to make another clip can be valuable. Someone can test a visual idea, explain a difficult concept or create something entertaining without a conventional production process. But the same capability can supply endless variations of an idea that barely deserved our time in the first place.

For a publisher, more output is an opportunity. For a viewer, it creates no additional time. That imbalance is why AI-generated short content deserves scrutiny: the supply can grow while the responsibility for deciding what is enough remains with the person watching.

Why the next video does not always satisfy

We often skip because the current clip is disappointing. It seems reasonable that a better one will make the experience more enjoyable. Research suggests that the act of repeatedly searching can sometimes work against that intention.

In seven experiments involving more than 1,200 participants, researchers Katy Tam and Michael Inzlicht examined digital switching and boredom. Participants expected switching to relieve boredom. In a video experiment, they instead reported less boredom and greater satisfaction, engagement and meaning when they stayed with one video.

The wider findings were less conclusive for some populations and for reading articles. Still, the video result suggests a useful possibility: we may keep seeking a more satisfying item while giving each one too little time to become satisfying.

When scrolling becomes automatic

A recent study developing the Binge Scrolling Scale gives this discussion more precise language. Published online in December 2025 and in the journal’s 2026 volume, it developed a 12-item measure covering three dimensions: automatic scrolling, negative consequences and loss of control.

The finding was a measurement tool: the researchers identified these three dimensions and tested the scale’s structure in separate samples of 262 and 233 adults. The study used convenience and snowball sampling in Turkey; it does not establish how common binge-scrolling is globally or prove that AI causes it. Its value here is the distinction between enjoying a long session and having difficulty controlling one. These dimensions are research concepts, not a self-diagnosis checklist.

What does this mean for our attention spans?

A 2025 meta-analysis in Psychological Bulletin combined 71 studies involving 98,299 participants across the full review. Greater short-form video use was associated with poorer cognitive outcomes, including attention and inhibitory control, the ability to suppress an impulse or response.

The pattern is concerning, but it leaves the direction of the relationship open. Attention difficulties could affect how people use short videos; viewing habits could also contribute to difficulties. Other factors may influence both. A June 2026 review of younger users illustrates the limitation: 88% of its 42 studies were cross-sectional, observing participants at one point in time.

Together, these studies suggest that attention deserves consideration alongside viewing time. A session can last a long time while our engagement with any individual item remains brief. For content creators, that raises a practical question: did we help someone understand or enjoy something, or merely keep them moving?

Short content can be worth our time

The pressure to earn attention predates generative AI. In 2011, Nielsen Norman Group reported that visitors often left web pages within 10–20 seconds, while pages that quickly communicated their value could hold attention much longer. The takeaway: visitors often decide quickly whether a page is worth staying on; a short visit does not establish a short capacity for attention. This is a historical finding, not a universal benchmark for today’s websites. For publishers, the implication is to make the value clear early, rather than assume every piece must be short.

A funny clip does not need to teach us anything to be worthwhile. A concise demonstration can save us a frustrating hour. An AI-generated animation can give an unfamiliar idea a form we can grasp.

The issue is whether the experience delivers what we wanted. A satisfying break and a session we repeatedly fail to end can look similar in a platform’s viewing figures. They feel very different to the person holding the phone.

For educational or commercial content, I would judge a short piece by what someone can take away from it. Does it answer the question it opens with? Does it retain the qualification that makes the answer accurate? Can someone find the original evidence when they need more detail?

These are content experience decisions. AI can help with the work, including turning a long explanation into something more approachable. As with AI’s role in independent work, the value depends on the result we choose to pursue, not simply the amount we produce.

Give yourself a place to stop

Changing the conditions of access may be more useful than repeatedly expecting ourselves to exercise better judgement in the middle of a feed.

A 2025 randomised trial tested a substantial change: blocking smartphone internet for two weeks while keeping calls, texts and internet access on other devices. After two weeks, the intervention group improved on an objective sustained-attention task; the comparison group, which still had mobile internet access, showed no significant change during that period. The study also shows how difficult a restriction can be to maintain.

A smaller change will not necessarily reproduce that result. But the trial gives us a reason to consider access and environment, alongside personal resolve. Start with a boundary that serves something you actually want: an undisturbed meal, a finished piece of work, or a break that ends when you intended.

AI will give us more ways to make content worth watching. It also gives us more ways to fill a feed. As creators, we can choose whether a piece leaves someone satisfied or simply waiting for the next instalment. As viewers, we can ask whether we are still choosing the next video, or have stopped making that decision.