Think about the last film you watched, album you listened to, meal you ate with friends or football match you sat through.
Did you make it to the end without checking your phone?
For many of us, the answer is becoming increasingly difficult.
The smartphone has inserted itself into the spaces between almost everything we do. A quiet moment becomes an opportunity to check a notification. A television advert becomes an opportunity to scroll. Waiting for a train becomes an opportunity to open an app. Even watching one screen increasingly happens while simultaneously looking at another.
This is not simply a story about people lacking self-control.
We have built a digital environment in which some of the world’s most sophisticated technology companies compete continuously for a finite human resource: our attention.
The uncomfortable question is what that competition is doing to us.
Britain Is Spending Four and a Half Hours a Day Online
Ofcom’s latest Online Nation research found that UK adults spent an average of four hours and 30 minutes online each day on personal smartphones, tablets and computers in May 2025.
Young adults spent considerably longer.
People aged 18 to 24 averaged six hours and 20 minutes online each day.
That does not mean those hours are automatically harmful. People work, learn, communicate, navigate, shop, bank, create, watch entertainment and maintain relationships online.
The internet provides enormous value.
The more interesting question is how much of that time we consciously intended to spend there.
There is an important difference between opening your phone because you have decided to do something and discovering 30 minutes later that you have been scrolling without remembering exactly why you started.
That difference is where digital wellbeing becomes a design question.
The End of the Page Changed Something
A newspaper eventually runs out of pages.
A television programme eventually reaches the credits.
An album eventually reaches its final track.
A social-media feed does not have to end.
Infinite scroll removed one of the natural stopping points that previously told us an activity was finished.
The significance of this design choice is easy to underestimate.
Stopping creates a decision.
When you reach the bottom of a page, you have to decide whether you want another one. When the next piece of content appears automatically, that decision can disappear.
Autoplay applies a similar principle to video. Notifications bring the application back into your attention after you have left it. Streaks can create a reason to return tomorrow. Personalised recommendations continuously provide another piece of content selected partly from signals about what previously captured your interest.
None of these mechanisms independently forces somebody to remain online.
Together, however, they can create environments with remarkably few natural opportunities to stop.
The Platform Has an Objective Too
When we use a social platform, two objectives can coexist.
The user may want entertainment, information, friendship or simply something to fill five minutes.
The platform may want engagement.
For advertising-supported businesses in particular, attention has economic value. More engagement can create more opportunities to show advertising, collect behavioural signals and learn what content keeps particular users interested.
Recommendation systems therefore analyse signals generated through our behaviour.
What did you watch?
What did you skip?
What did you like?
What did you share?
What did you comment on?
How long did you stay?
What did you watch next?
Those signals help determine what the system believes might keep you interested in the future.
The feed gradually becomes personalised.
But personalised does not necessarily mean representative.
Your Feed Is Not the World
This distinction matters enormously.
Imagine watching several videos about a particular political controversy.
The recommendation system observes your behaviour and may conclude that you are interested in that subject. It can therefore recommend additional content related to it.
You watch more.
The system receives another signal.
More related content becomes available.
Over time, the digital environment presented to you can become increasingly different from the environment presented to somebody sitting beside you.
Neither feed is necessarily a neutral representation of reality.
Each is a computational selection.
This is why two people can open the same platform and encounter radically different versions of what appears to be happening in the world.
Emotion Complicates the Picture
Emotionally charged information can perform particularly well online.
Research has repeatedly found relationships between moral-emotional language, anger, outrage, group identity and online sharing in certain contexts, particularly political communication.
This does not mean that algorithms simply contain an instruction saying “show people angry content”.
Nor does it mean that negative content always outperforms positive content.
The mechanisms are considerably more complicated.
A new 2026 study in Nature Human Behaviour, for example, suggests there may even be a saturation point at which adding more moral language predicts lower engagement.
What the wider evidence does suggest is that recommendation systems operating on engagement signals can interact with human psychological tendencies in ways that amplify emotionally powerful material.
If outrage generates comments, replies, shares and prolonged attention, a system optimising engagement can learn that the content is effective without understanding the emotional consequences of repeatedly presenting it.
The algorithm does not need to feel anger.
It only needs to measure behaviour.
The Feedback Loop Matters
This can create a powerful feedback loop.
A user engages with emotionally charged content. The system interprets that engagement as evidence of interest. Related material becomes more likely to appear. Creators observe which posts perform successfully and produce more material capable of generating similar engagement.
Users, creators and recommendation systems begin influencing one another.
The result does not have to be an inevitable descent into extremism or anxiety. People respond differently, platforms use different recommendation systems, and users actively shape their own feeds through countless choices.
Nevertheless, the architecture matters.
Research published in Nature Human Behaviour has examined how emotion can become amplified through social-media environments, while other studies have shown that anger and disgust can reinforce one another during online interaction.
Digital environments do not simply contain human behaviour.
Their architecture can shape it.
Are We Chasing the Next Reward?
There is an old expression associated with repeatedly pursuing an elusive high: “chasing the dragon”.
The comparison should be used carefully because compulsive social-media use is not equivalent to substance dependence.
But the metaphor captures something recognisable about scrolling.
You are not necessarily searching for one specific thing.
You are searching for the next interesting thing.
The next joke.
The next goal.
The next argument.
The next piece of gossip.
The next breaking story.
The next surprising fact.
The next video that teaches you something.
Most of what appears may be forgettable.
But occasionally something is extremely rewarding.
That uncertainty itself can keep us moving.
One more swipe might contain something better.
The Cost May Be Fragmented Attention
The important question is therefore not simply how many hours we spend online.
It is what happens to the rest of our lives when our attention becomes continuously interruptible.
A film competes with WhatsApp.
A conversation competes with Instagram.
A football match competes with notifications.
Reading competes with TikTok.
Work competes with everything.
Even boredom now competes with the smartphone.
That last point may be particularly important because boredom historically created space for daydreaming, reflection, observation and spontaneous conversation.
We have developed a device capable of eliminating almost every empty moment.
We have not yet fully understood what happens when we do.
Mental Health Requires More Careful Language
The relationship between social-media use and mental health is important, but it should not be reduced to the claim that social media automatically causes anxiety or depression.
Research is considerably more nuanced.
Associations have been found between problematic or intensive social-media use and outcomes including poorer sleep, anxiety, depression and wellbeing, particularly among some young people.
But individuals differ significantly. The type of activity matters. The content matters. Existing vulnerabilities matter. Online communities can also provide friendship, information and support.
The better question is therefore not whether the internet is good or bad for mental health.
It is which combinations of design, behaviour, content and individual circumstances create benefits or harms.
That question is much harder.
It is also much more useful.
Why Does Responsibility Sit Almost Entirely With the User?
Digital wellbeing advice frequently focuses on individual behaviour.
Turn off your notifications.
Set a screen-time limit.
Put your phone in another room.
Take a digital detox.
Those interventions can be useful.
But they create an interesting asymmetry.
Some platforms employ enormous teams of software engineers, behavioural researchers, data scientists, designers and machine-learning specialists to understand user behaviour.
The individual is then expected to counter that environment primarily through willpower.
This does not remove personal responsibility.
It raises the question of whether responsibility should be shared.
If particular design choices demonstrably encourage excessive or compulsive use, digital wellbeing should become a product-design consideration rather than merely a user-discipline problem.
What Would Technology Designed for Wellbeing Look Like?
The next generation of digital products could optimise for more than engagement.
| Design Intervention | What It Could Change |
|---|---|
| Natural stopping points | Give users moments to consciously decide whether they want to continue. |
| Optional chronological feeds | Reduce dependence on engagement-based ranking where users prefer a simpler experience. |
| Notification controls | Make interruption something users deliberately configure rather than accept by default. |
| Recommendation explanations | Show why particular content is appearing and which behaviours influenced the recommendation. |
| Algorithmic choice | Allow people to select between different recommendation objectives. |
| Wellbeing signals | Incorporate user-defined measures of satisfaction rather than treating continued engagement as the primary success signal. |
| Intensity controls | Allow users to reduce repetitive recommendations around emotionally charged subjects. |
| Attention dashboards | Show not only screen time but which systems repeatedly captured a person’s attention. |
The interesting innovation opportunity is not necessarily another application telling people they have spent too much time on their phone.
The operating systems already provide versions of that information.
The bigger opportunity is designing digital environments that respect intentional attention in the first place.
Could We Have a Right to Choose Our Algorithm?
This leads to a much bigger policy and technology question.
Users can choose which newspaper to buy.
They can choose which television channel to watch.
They can choose which podcast to hear.
But once somebody enters many modern social platforms, the ranking system deciding what appears next is largely determined by the platform.
What if algorithmic choice became a normal product feature?
A user might choose a chronological feed.
Another might choose recommendations optimised for learning.
Someone else might prioritise content from people they actually know.
Another person might deliberately reduce political content or highly repetitive recommendations.
An algorithm would still be involved in many of these experiences.
The difference would be that the objective becomes partly chosen by the person whose attention is being allocated.
That could represent a significant shift in digital power.
The UKBT Institute Opportunity
This is exactly why digital wellbeing cannot be treated purely as a computer-science problem.
Computer scientists can explain recommendation systems.
Psychologists can investigate attention and behaviour.
Neuroscientists can study cognition.
Designers can create different interfaces.
Teachers can observe how technology affects learning environments.
Sociologists can examine how platforms reshape communities.
Policymakers can consider appropriate protections.
Young people can tell us what living inside these systems actually feels like.
The challenge requires those disciplines to meet.
The UKBT Institute’s Living in a Digital World campaign is interested in precisely these questions because the future of technology should not simply be about making systems more powerful.
It should also involve deciding what kind of relationship we want those systems to have with human beings.
The Bigger Picture
The smartphone is one of the most useful technologies ever created.
It is a camera, map, bank, library, television, newspaper, music player, workplace, classroom and communications network sitting in a pocket.
The problem is not that we use it.
The question is whether we always remain the ones deciding when and how that use happens.
Digital environments increasingly learn what captures our attention, predict what we might engage with next and personalise themselves around those signals.
That capability can create extraordinary convenience.
It can also make disengagement increasingly difficult.
As technology becomes better at predicting what will make us look, click, watch and respond, digital wellbeing will require more than telling individuals to develop stronger self-control.
We will also need to ask what responsibilities belong to the people designing the environments in which that self-control is being exercised.
The next time you finish reading something online, resist the immediate swipe for just a moment. Notice the urge to find out what comes next, and ask yourself whether continuing is a decision you have consciously made or simply the default the interface has made easy.
Your feed does not show you the world. It shows you a selection of the world shaped partly by what the system has learned can hold your attention. The next digital right worth debating may be the right to decide how that selection is made.