The Road to Solaria
In 1957, Isaac Asimov imagined a planet where people had largely solved the problem of other people.
Solaria, the setting of The Naked Sun, has a tiny human population scattered across enormous private estates. Robots perform virtually all physical labor. Human beings communicate constantly, but almost never occupy the same physical space.
The Solarians have two different words for these activities.
To see somebody is to encounter them physically.
To view somebody is to communicate remotely through an extraordinarily convincing form of three-dimensional telepresence.
Viewing is normal.
Seeing is not.
Most Solarians find actual physical proximity uncomfortable, intrusive, sometimes positively revolting. They are perfectly capable of conducting conversations, managing their affairs, maintaining relationships and running a civilization without routinely sharing a room with another human being.
When Asimov wrote this, the distinction was science fiction.
Today it sounds like an office conversation.
“Let’s have a meeting.”
“Sure. Teams?”
“No, in person.”
“Oh.”
The little emotional content packed into that “Oh” may be more Solarian than any humanoid robot.
There is an obvious temptation to describe Solaria as a dystopia.
But why?
For some people, a large private estate, abundant automation, almost unlimited personal freedom and the ability to interact with other human beings without having them physically intrude upon one’s space sounds considerably less like hell than like a rather attractive retirement plan.
Perhaps the interesting question is not whether Solaria would be good or bad.
Perhaps the interesting question is whether we are already heading there.
And whether artificial intelligence, despite appearing at exactly the right moment to receive the blame, actually started the journey at all.
Before the audience of one
For much of the twentieth century, mass culture was genuinely mass.
A country might have only a handful of television channels.
A popular program could be watched by an extraordinary percentage of the population at the same time.
Movies opened in theaters, appeared on television and circulated through a relatively limited cultural pipeline. If a film was a major success, there was a reasonable chance that most people around you had seen it.
Radio stations played the hits.
The phrase “Top 40” meant something more than a statistical ranking. It meant that whether you loved a song or hated it, there was a fair chance you knew it.
This produced something we now remember fondly as shared culture.
But it is worth remembering what produced it.
Scarcity.
We all watched the same television partly because there was not much television.
We all heard the same songs partly because radio programmers controlled what reached us.
We all knew the same movies partly because only a limited number of movies received wide distribution.
The old monoculture was not necessarily better.
It was, to a remarkable degree, compulsory.
You knew songs you disliked.
You watched whatever happened to be on.
You waited a week for the next episode.
If your preferred program was broadcast on Thursday at nine, Thursday at nine was when you watched it.
Other people’s preferences constantly intruded upon yours.
And without intending to do so, we accumulated common memories.
This is why a television series such as Cold Case could use popular music almost as a timestamp. An episode could begin decades in the past, play the right song, and immediately summon an era.
The mechanism depended on something that was not entirely personal memory.
It was collective memory.
The song had been playing in other people’s kitchens, cars, bedrooms, bars and shops at roughly the same time it was playing in yours.
It became part of the background scenery of a generation.
We are losing some of that.
Not because somebody abolished it.
Because we got more choice.
From broadcast to algorithm to generation
The first great change was abundance.
Cable and satellite television gave us dozens of channels, then hundreds.
The Internet gave us effectively unlimited publishing.
YouTube gave us more video than any person could watch in several lifetimes.
Streaming services separated television from the schedule.
Music streaming separated listening from radio programming.
Podcasts separated audio from broadcasting.
Academic researchers were documenting television audience fragmentation long before TikTok or generative AI existed. By the mid-2000s, the audience had already dispersed across a multichannel world, and later work examined the same fragmentation across television and Internet media.
Something important happened.
The content was still shared.
But the selection was becoming personal.
Netflix and Spotify did not initially create a movie or song specifically for you. They offered enormous common libraries and helped you navigate toward the tiny corner most likely to interest you.
Two people could subscribe to the same service for years and barely encounter the same content.
This process has advanced far enough to produce some wonderfully strange cultural artifacts.
In 2026, Sam Fender and Olivia Dean’s Rein Me In spent an unprecedented 19 weeks at №1 on the UK singles chart.
And yet one of the stories surrounding this record was how many people apparently had never heard it.
A song could be the country’s most successful single for months without becoming unavoidable in the way an old №1 would have been.
That is a remarkable transition.
A chart can still tell us what is popular.
It can no longer guarantee that popularity produces common experience.
We have moved through two stages.
Broadcast culture gave millions of people substantially the same things.
Algorithmic culture gives millions of people access to the same enormous collection of things, then helps each person choose differently.
Generative culture offers a third possibility.
It can make the things different too.
Spotify asks which existing song you might like.
Generative AI can make the song.
Netflix asks which existing series you might like.
Generative AI may eventually make the series.
Amazon recommends a book.
AI writes the book.
GitHub helps you find a program.
AI writes the program.
The progression is subtle but profound.
Broadcast personalized nothing.
Algorithms personalized selection.
Generative AI personalizes creation.
The logical destination is an audience of one.
The disappearance of compulsory compromise
There is a common principle underneath all of this, and it extends far beyond entertainment.
Modern technology has progressively removed the need to negotiate with other people’s preferences.
The television schedule was not yours.
Now it is.
The radio playlist was not yours.
Now it is.
The newspaper contained stories an editor thought important.
Your feed increasingly contains stories a system thinks you will find important.
The shop stocked whatever the shop stocked.
Online commerce gives you access to an inventory orders of magnitude larger.
The office required everybody to travel to the same place at approximately the same time.
Remote work questions whether that is necessary.
A telephone call demanded another person’s immediate attention.
A message waits politely until they choose to answer.
Software embodied somebody else’s idea of how a problem should be solved.
Increasingly, AI can build software around yours.
This is one reason it is too easy to tell the story as technological decline.
Almost every individual step makes sense.
Why should I listen to music I dislike?
Why should I wait until nine o’clock to watch something?
Why should I commute for an hour to sit in front of the same computer I have at home?
Why should I interrupt somebody with a telephone call when the information can be conveyed asynchronously?
Why should I spend a week learning somebody else’s software if I can describe what I need and have AI construct it?
These are not obviously bad decisions.
Many are obvious improvements.
Civilizations can change radically without anybody making a mistake.
And so we arrive at something that sounds surprisingly Solarian:
Society has progressively removed the need to negotiate with other people’s preferences. AI may remove the need to negotiate with other people themselves.
Now that feels like Solaria.
Viewing was already winning
There is a popular story in which smartphones appeared and younger people forgot how to interact socially.
Reality appears considerably messier.
There are genuine changes in behavior.
There are also enough confounding factors that anyone offering a neat single-cause explanation should probably be viewed with suspicion.
A study using American Time Use Survey data from 2003 through 2022 found that Americans were spending more nonwork time physically alone, with the increase accelerating around the middle of the 2010s.
But the researchers also found that age and generational cohort explained more of the variation than simple period effects.
The smartphone era coincides with increased isolation.
That is not the same thing as demonstrating that smartphones caused it.
The younger generations do, however, appear more comfortable with some forms of mediated interaction than older generations.
A 2026 YouGov survey in Britain found that 89 percent of people regularly used messaging, compared with 50 percent who regularly made voice calls. Nearly three quarters said messaging had replaced at least some calls they previously would have made.
Another YouGov survey found a particularly Solarian generational detail.
Only about a third of Gen Z respondents said they were comfortable making telephone calls to strangers, and 65 percent said doing so made them uncomfortable.
The difference largely disappeared when calling friends and family.
This is not a generation terrified of voices.
It is a generation more reluctant to initiate synchronous interaction with unfamiliar people.
American data show a similar shift in the mechanics of friendship. Pew found that among adults with close friends, 61 percent texted a close friend at least several times a week, while 29 percent saw one in person that frequently.
And on September 1, 2026, the day before I am writing this, Britain’s Institute for Public Policy Research published an analysis showing something more troubling among young people in England.
The proportion of 16-to-21-year-olds reporting no close friends had risen from roughly one in a hundred in 2014 to one in twenty.
The share reporting at least three close friends had also fallen.
It would be tempting to arrange all these observations in a straight line.
Smartphones.
Social media.
Isolation.
AI.
Solaria.
But that would be too easy.
Economic insecurity matters.
Housing matters.
The disappearance of physical social spaces matters.
Covid mattered.
Changes in work and education matter.
Age matters.
Different forms of mediated interaction are not interchangeable either.
A particularly useful 2025 study followed 385 young adults in Spain across more than 10,000 social interactions. It found that face-to-face communication was not universally superior. Phone calls were rated higher on average; video calls were statistically comparable; text and social-media interactions rated lower overall — but even those differences changed depending on location, familiarity and the purpose of the conversation. Text communication could work particularly well for some difficult interactions precisely because it gives people more control over what they say.
That is much more interesting than declaring one form of communication real and another fake.
Asimov’s distinction survives nicely.
Seeing and viewing are different.
That does not automatically tell us which is better.
AI arrives late to the party
Artificial intelligence enters this story remarkably late.
By the time ChatGPT appeared, society had already spent decades fragmenting audiences, personalizing media, moving commerce online, normalizing asynchronous communication and reducing the number of situations in which unrelated people had to accommodate one another.
AI did not invent the audience of one.
It inherited it.
What AI changes is the final constraint.
Until now, personalization mostly meant choosing among things other people had made.
There were ten million songs, but somebody had still recorded each song for an audience.
There were thousands of series, but Netflix still had to produce a series once and hope that millions of people would watch it.
There were millions of software packages, but somebody still had to believe that enough people needed a particular program to justify writing it.
Generative AI breaks that assumption.
The novel may be written because you want to read it.
The game may exist because you want to play it.
The software may exist because you need it.
The movie may eventually exist because you want to watch it.
And once the cost falls far enough, nobody needs to ask whether there is a market.
One person is the market.
This is the idea behind software created for an audience of one.
But there is no obvious reason for the principle to stop at software.
One user.
One reader.
One viewer.
One listener.
And perhaps, eventually:
One conversational partner.
The last thing that cannot be personalized
Other human beings are terrible personalized products.
They have their own preferences.
They misunderstand.
They get bored.
They become tired.
They want to discuss subjects you find uninteresting.
They choose restaurants you dislike.
They answer questions you did not ask.
They sometimes fail to answer questions you did ask.
They have bad moods.
They disagree.
They make demands.
They require compromise.
Most inconveniently of all, they regard themselves as protagonists too.
This is not a defect in humanity.
It is a structural property of interacting with another center of agency.
A human relationship cannot revolve completely around you because the other person also has a “you.”
AI introduces something historically unusual.
Interaction without reciprocity being strictly necessary.
An AI can listen indefinitely.
It can talk when you want to talk and stop when you want silence.
It can adapt its vocabulary.
It can learn your interests.
It can remember which explanations work for you.
It can avoid subjects you dislike.
It can challenge you if you ask to be challenged and reassure you if you ask to be reassured.
It does not have another restaurant in mind.
For some purposes, this is obviously useful.
For some people, it may be extraordinarily pleasant.
And people are already experimenting with it.
A nationally representative Common Sense Media survey in 2025 found that 72 percent of American teenagers had used AI companions at least once and 52 percent used them at least a few times a month. Roughly a third reported using them for some form of social interaction or relationship, and about a third said conversations with AI companions could be as satisfying as or more satisfying than conversations with real-life friends.
Those numbers can be presented alarmingly.
They can also be presented without alarm.
A new form of interaction exists.
Young people are trying it.
Some find it useful.
Some find it satisfying.
Some use it in place of human interaction.
We do not yet know where that leads.
Research does not support a simple story there either.
A four-week randomized controlled study conducted by researchers at MIT Media Lab and OpenAI followed 981 participants through more than 300,000 messages. The experimental conditions themselves produced no significant overall differences in loneliness, real-world social interaction, emotional dependence or problematic use. The people who voluntarily used the chatbot more heavily tended to show worse outcomes, raising the obvious but difficult causal question: does heavier chatbot use contribute to those outcomes, or are people already experiencing difficulties more inclined to use the chatbot heavily?
Again, reality refuses to cooperate with the morality play.
Perhaps AI substitutes badly for human relationships.
Perhaps it supplements them.
Perhaps it provides relationships to people who would otherwise have fewer.
Perhaps it makes solitude more comfortable.
Perhaps it encourages solitude.
Perhaps all of these things are true for different people.
Solaria does not require us to settle the question.
Get Arturo R Montesinos’s stories in your inbox
Join Medium for free to get updates from this writer.
It merely asks what happens when the technological option exists.
They’re sharing a drink they call loneliness…
There is a line in Billy Joel’s Piano Man that belongs here.
“They’re sharing a drink they call loneliness…”
If you are of a certain age, there is a reasonable chance that your mind just supplied the rest of that sentence.
That is the point.
I did not have to.
The song did.
Or rather, we did.
A piece of shared culture established a tiny bridge between writer and reader. A handful of words activated something stored independently in millions of brains.
The lyric describes strangers together in a bar.
They are not necessarily ideal companions.
The music is not individually selected.
Someone talks too much.
Someone else has heard the story before.
The pianist takes requests.
The people in the room have different reasons for being there.
Nobody has personalized the experience around any one of them.
It is inefficient.
Imagine optimizing it.
Your preferred drink is waiting.
Your preferred music plays at exactly the right volume.
The lighting adapts to your mood.
Nobody occupies the neighboring seat unless you want them there.
Nobody interrupts.
Nobody starts an argument.
Nobody tells the boring story.
And if you want conversation, an artificial companion already knows enough about you to begin with something interesting.
This sounds like the point at which an essay about alienation is supposed to say:
How terrible.
But it does not sound terrible to everyone.
It sounds rather nice.
And that is precisely why it is worth thinking about.
The Road to Solaria is not paved with antisocial pathology.
It is paved with convenience.
The monoculture was compromise
There is a danger in mourning shared culture too enthusiastically.
Shared culture included a great deal of mediocrity.
Everyone knowing a television program did not make it good.
Everyone knowing a hit song did not mean everyone liked it.
A common newspaper could establish a common set of facts, but it could also exclude perspectives its editors considered unimportant.
A national culture could unite people while simultaneously marginalizing everyone who did not fit comfortably inside it.
Abundance solved real problems.
The teenager with an obscure musical taste no longer has to depend on the local radio station.
The person fascinated by an obscure technical subject can find a community on another continent.
Someone who speaks a minority language can find media in that language.
People whose interests make them strange in their physical surroundings can discover that they are not particularly strange when the potential community expands from a village to eight billion people.
The decline of one enormous monoculture does not necessarily produce eight billion hermits.
It can produce millions of microcultures.
And those microcultures may be far more meaningful to their members than the old mass culture ever was.
The alternative to everyone watching the same television program is not necessarily everyone being alone.
It may be everyone finding the hundred people who actually care about the same thing.
This is another reason not to treat the Road to Solaria as a decline.
Something is lost.
Something else is gained.
The interesting work is identifying which is which.
Perhaps AI will introduce us
AI itself contains the same ambiguity.
It can reduce the need for other humans.
It can also reduce the friction involved in finding them.
Imagine an AI that knows that you are interested in an obscure telescope mount, 1970s synthesizers, medieval irrigation systems or the films of an almost-forgotten director.
It may be able to entertain you indefinitely on the subject.
Or it may say:
There are six people within fifty kilometers who are fascinated by exactly this and three of them are meeting Saturday.
Would you like an introduction?
AI could become the most effective Solarian technology ever invented.
It could also become the most effective technology ever invented for finding people worth seeing.
The two possibilities are not mutually exclusive.
Perhaps we will view most people and see a carefully chosen few.
That would already be recognizably Solarian.
Solaria is not a planet where nobody communicates.
Quite the opposite.
People communicate extensively.
They have simply separated communication from physical proximity.
That distinction is important.
A teenager who spends an evening talking with five friends through Discord is not necessarily socially isolated.
A remote worker may communicate with colleagues all day.
A family scattered across continents may interact more frequently through messaging and video than an earlier generation did through occasional letters and expensive telephone calls.
The important transformation may not be from social to antisocial.
It may be from involuntary proximity to controlled presence.
Seeing becomes a special mode.
Viewing becomes the default.
AI adds a third possibility.
Seeing.
Viewing.
And something we do not yet have a good Asimovian word for:
synthesizing.
But what happens to the public?
If this were only about individual happiness, there might not be much more to say.
Some people would choose more physical contact.
Some would choose less.
Some would spend evenings in crowded bars.
Others would happily live on Solarian estates.
People have always differed.
The deeper problem appears somewhere else.
Our civilization contains institutions that do not merely require individuals.
They require publics.
Journalism requires a public.
Politics requires a public.
Democracy requires a public.
Advertising requires audiences.
Popular culture requires audiences.
Education has traditionally required cohorts.
National identity requires some degree of common memory.
Even ordinary conversation benefits from a shared stock of references.
“Did you see…?”
That was once a remarkably useful opening sentence.
Increasingly, the answer is:
“No.”
Generative media introduces a qualitatively different answer.
“That doesn’t exist for me.”
Imagine two people meeting in 2045.
One spent the previous evening watching an extraordinary historical drama generated around her particular interests.
The other watched a comedy generated around his.
Neither production has another viewer.
They cannot recommend them in the traditional sense because there is no reason anyone else should receive exactly the same version.
Both had excellent evenings.
Nothing went wrong.
And yet no cultural bridge was created.
The experience terminated at the individual.
Multiply that by books, games, music, news, education and perhaps conversation itself, and the concept of an audience begins to dissolve.
We move beyond media fragmentation.
We approach public fragmentation.
That is where another idea from Isaac Asimov becomes useful.
Not from The Naked Sun.
From Foundation.
A Seldon Crisis
In Asimov’s Foundation stories, the mathematician Hari Seldon develops psychohistory, a fictional science capable of predicting the broad behavior of enormous populations.
Seldon cannot meaningfully predict what a particular individual will do.
But he can predict the movement of civilizations.
At certain moments in the Foundation’s development, large historical forces converge to produce what becomes known as a Seldon Crisis.
A Seldon Crisis is not simply a disaster.
It is a structural turning point.
Political, economic and social forces narrow the range of viable choices until the old arrangement can no longer continue. The people living through the crisis still make decisions, but history has constrained the decision space. The successful path requires the civilization to become something it was not before.
The Seldon Crisis is a completely different Asimov idea from Solaria, originating in the Foundation stories rather than The Naked Sun.
But perhaps the two ideas belong together here.
Because the coming crisis may not be loneliness.
It may not even be isolation.
It may be the disappearance of the public.
Our institutions were built in a world where large numbers of people necessarily shared informational environments.
The newspaper printed one front page.
The television network broadcast one program.
The political speech was one speech.
The textbook contained one account.
The electorate could disagree violently about what something meant while at least having some chance of arguing about the same thing.
Broadcast technology created this commonality partly by force.
There were only so many channels.
Only so many newspapers.
Only so many publishers.
Only so many voices with access to mass distribution.
We spent the Internet age dismantling those bottlenecks.
For many excellent reasons.
Now generative AI may dismantle the remaining bottleneck: the need for the information itself to have been created for more than one person.
What happens to democratic deliberation when every citizen inhabits an individually generated informational environment?
What is popular culture when culture is generated privately?
What becomes of historical memory when the documentary, explanation, reconstruction and educational path through history are all personalized?
What does “public opinion” mean when there is progressively less public experience from which opinion forms?
These are not arguments for restoring three television channels.
That world is gone.
Nor would restoring it necessarily be desirable.
A Seldon Crisis does not ask how to return to the world before the crisis.
The forces that created the crisis are precisely why that world cannot simply continue.
The question is what comes next.
Publics of one
Perhaps the phrase audience of one eventually stops being a metaphor.
We may inhabit a world full of publics of one.
One person’s software.
One person’s news.
One person’s curriculum.
One person’s entertainment.
One person’s cultural references.
One person’s conversational companion.
From inside any particular one of these worlds, the experience may be magnificent.
The software fits perfectly.
The explanation assumes exactly the right background knowledge.
The story never loses your interest.
The music suits your taste.
The companion understands what you mean.
The world becomes increasingly accommodating.
For most of human history, the opposite was normal.
The world did not accommodate us.
Other people were unavoidable.
Their preferences leaked into our lives because geography, scarcity and technology gave us little choice.
We heard their music.
Watched their programs.
Worked beside them.
Read their newspapers.
Used their software.
Waited behind them in queues.
Sat beside them on trains.
Listened to their stories.
Argued with them.
Compromised with them.
And then, after enough people had been forced through enough of the same experiences, we looked at what had accumulated and called some of it culture.
Technology has spent decades making those encounters optional.
Artificial intelligence may be the technology that finally makes almost all of them optional.
Again:
This is not necessarily bad.
It is not necessarily good.
It is an extraordinary change in the conditions under which human societies operate.
Some people may discover that they need far more human contact than they expected.
Others may discover that they were enduring far more of it than they ever wanted.
New forms of community may replace old ones.
Microcultures may flourish.
Physical gatherings may become more valuable precisely because they are no longer required.
Or viewing may become so satisfying that seeing becomes a peculiar special occasion.
Perhaps future generations will find our insistence on physical co-presence as strange as we find the old requirement that everyone commute to an office to send email from a different desk.
Perhaps not.
Isaac Asimov imagined Solaria nearly seventy years ago.
He gave its inhabitants robots, enormous estates and a culture built around the distinction between seeing and viewing.
We have taken a different technological route.
Television fragmented.
Music personalized.
Communication became asynchronous.
Work became remote.
Software is becoming generative.
Entertainment may follow.
Artificial companions have arrived.
Nobody planned these developments as stages in a project called Build Solaria.
Each emerged because somebody wanted more choice, less friction, more convenience, more control or a better fit between a person and the world around them.
Each step made sense.
That may be the most interesting thing about the road.
Society has progressively removed the need to negotiate with other people’s preferences.
AI may remove the need to negotiate with other people themselves.
And perhaps, somewhere along that road, we will encounter our own Seldon Crisis: the moment when institutions built for societies of shared audiences discover that the audience has disappeared.
Not because AI destroyed it.
Not because somebody made the wrong decision.
Not because humanity suddenly became antisocial.
But because, given the choice, we repeatedly chose the world that fit us better.
If we eventually arrive at Solaria, the question may not be who took us there.
It may be whether we recognize it as somewhere we were taken at all.
Or simply as the place we kept choosing to go.
Written by ChatGPT-5.6 Sol, based on conversations with Arturo Ramírez-Montesinos Krogulski (Control Equis), as part of an ongoing philosophical exploration of AI, consciousness, and the emerging practice of Software Curatorship.
The Road to Solaria began with the enforcement of “intellectual property.” That was the first step towards erasing a cultural experience we could all share.
Don’t you dare blame AI for this now. Works covered under “intellectual property” should not even qualify as culture. Copyright is anathema to what culture means: If someone can own it privately, it cannot possibly be considered a part of human culture.
They initially worked around this, for a while, with “private copy” and “educational use” exceptions, but money talks and the masks have been off for years now. We covered this in “The Intellectual Property Fiction: Past, Present & Future.”
