A hundred years is a useful distance. Long enough that the current news cycle is ancient history, short enough that some people alive in 2126 will have living memory of people who were alive in 2026. The children being born this week have a non-trivial chance of being interviewed, in their late nineties, about what the early AI era was actually like. That matters. It makes the 100-year horizon a question about the world people we know will inherit, not an abstract one.

I wrote recently about the next decade of AI and, at the other extreme, about what a thousand years from now might look like. This is the post that sits in the middle - the horizon where you can still be specific without being silly, and where the current generation’s decisions are still visibly reflected in the outcome.

The 1926 Calibration

Someone in 1926 trying to predict 2026 would have had partial success. They could have predicted commercial air travel, because Lindbergh’s Atlantic crossing was already on the horizon. They could have predicted television, because the underlying technology was being demonstrated that year. They would have got radio, antibiotics, the automobile’s dominance of cities, and the continued rise of the United States as an economic superpower.

They would have missed almost everything else. They did not predict the computer, let alone the internet. They did not predict the collapse of colonial empires, the Cold War, or the rapid rise of East Asia. They did not predict the contraceptive pill or the social revolutions it helped enable. They did not predict climate change as a mainstream concern, though some scientists were already noticing carbon dioxide levels. They did not predict that the average human would be dramatically taller, live decades longer, and spend most of their waking hours staring at a lit rectangle.

The pattern is that the hundred-year horizon is close enough that some current trends play out roughly as expected, and far enough that the really disruptive things come from directions nobody was looking in. A confident prediction of 2126 will probably be half-right, and the interesting half will be the part we cannot see from here.

With that caveat stated loudly, here is what I think the honest, probability-weighted picture actually looks like.

A Rough Timeline

 Year   Milestone
 ────   ──────────────────────────────────────────────────────
 2026 ──┬── You are here. AI era begins in earnest.
        │
 2035 ──┼── Routine cognitive work largely automated.
        │
 2045 ──┼── Clean energy cost approaches delivery cost.
        │       Fusion demonstrably commercial.
        │
 2055 ──┼── First generation with no pre-AI memory hits middle age.
        │       Gigatonne-scale carbon removal operational.
        │
 2065 ──┼── Global population peaks near 10B, decline begins.
        │       Off-world workforce in the thousands.
        │
 2075 ──┼── Brain-computer interfaces mainstream medical devices.
        │       Longevity interventions compound: +10 healthy years.
        │
 2085 ──┼── Pension systems either transformed or abandoned.
        │       Asteroid mining and orbital manufacturing routine.
        │
 2095 ──┼── Cities that grew for 200 years are visibly shrinking.
        │       Moral-status debate for AI enters legal systems.
        │
 2105 ──┼── Climate: stabilised below 3C, or in partial retreat.
        │       The fork is now visible in hindsight.
        │
 2115 ──┼── Stewardship-of-continuity profession matures.
        │       Tens of thousands live off Earth full-time.
        │
 2126 ──┴── 100-year-olds are ordinary. Children are rare and valued.
            The generation born in 2026 is being interviewed.

The milestone dates are directional, not precise. The ordering is the part that matters: energy and automation resolve first, demography and longevity in the middle, institutional and off-world shifts last.

What Is Almost Certainly Solved

Some problems that dominate the current decade will be historical footnotes by 2126.

Large-scale clean energy is the most important of these. Between solar, wind, grid-scale storage, and at least one if not both of fusion and advanced fission, the cost of electricity at the generation side approaches the cost of the infrastructure needed to deliver it. Energy as a binding constraint on what civilisation can do effectively ends within the next fifty years, and by 2126 we are living in the post-energy-scarcity economy that the twentieth century kept almost arriving at.

Most infectious disease is genuinely solved, not in the sense that nobody gets sick but in the sense that the ability to characterise a pathogen and deploy a therapy within days rather than months is mature. mRNA platforms, and whatever succeeds them, have been refined through a century of use. The 2020 pandemic looks, from 2126, the way the 1918 flu looks from 2026 - a serious event, handled imperfectly, remembered but no longer defining.

Routine cognitive work is fully automated. The debate about whether AI will take over knowledge work is, from the 2126 perspective, as quaint as the early twentieth-century debate about whether the automobile would replace the horse. It did. The replacement was uneven, messy, and disruptive for a generation. Then it was normal, and then it was invisible, and then a generation later it was the only thing that had ever made sense. The 2126 equivalent of a 2026 knowledge worker does something that takes the automation as a given and builds on top of it.

Longevity medicine works, at least partially. I am more cautious here than the aggressive longevity advocates, but the compounding effect of incremental gains in metabolic health, inflammation control, and cellular repair across a century is substantial. A 2126 human in a wealthy country probably has a healthy life expectancy twenty to thirty years longer than their 2026 counterpart. A few of the more aggressive interventions - probably senolytics, probably some form of cellular reprogramming, possibly something none of us can currently name - will have worked. Ageing as an inevitable, untreatable process will have been retired from the medical textbook.

What Is Almost Certainly Not Solved

Climate is the hardest one, and I want to be careful here because it is the problem most shaped by what we do in the next thirty years rather than the next hundred.

The temperature trajectory from 1.2C of warming to somewhere between 2C and 3C is now largely locked in regardless of what happens next, because of the thermal inertia of the oceans and the carbon already in the atmosphere. By 2126, humanity has either stabilised below 3C through a combination of rapid decarbonisation and large-scale carbon removal, or it has not. In the former case, the climate is a managed problem with visible scars - lost coastlines, rewilded abandoned zones, transformed agriculture - but no longer an existential one. In the latter case, 2126 is a civilisation in partial retreat, organised heavily around adaptation rather than progress.

My honest read is that we end up somewhere in the middle, leaning toward the managed case. The economics of clean energy have shifted faster than anyone predicted. The political will has not, but it doesn’t need to keep up with the economics for decarbonisation to happen; it just needs to not actively block it. Carbon removal at gigatonne scale by 2126 is extremely likely, because the incentives will compound once a credible market exists. But the climate that 2126 inherits is meaningfully hotter than the one we grew up in, and a lot of the geography has moved.

Geopolitical fragmentation is also probably not solved. The dream that technology produces a single global civilisation has been around since the nineteenth century and keeps not happening. By 2126 the specific alignments will be unrecognisable - it is very unlikely that the US-China axis still defines global politics, and it is likelier than not that at least one current great power has split or been replaced - but the underlying pattern of multiple competing spheres persists. People are tribal, and ambient intelligence does not automatically fix that.

Meaning, purpose, and mental health are not solved. In fact, I expect them to be meaningfully harder in 2126 than in 2026, for the simple reason that a civilisation which has automated most of the work its grandparents did for identity has to find a new answer to “what is a life for?” The early signs of that crisis are visible already. A century of compounding does not make it easier.

The Demographic Inversion

One of the most structurally important facts about 2126 is that there are probably fewer humans alive then than there are in 2050.

The UN’s population projections show global population peaking somewhere between 2060 and 2080 and then declining. That decline, once it starts, is stubborn, because it is driven by fertility rates that have fallen below replacement in almost every country that has modernised, and have stayed there. Subsidies and pro-natal policies have a small effect. Culture and economics have a much larger one, and the economics are not cooperating.

By 2126, the world population is probably somewhere between six and nine billion, down from a peak above ten. The age distribution is radically inverted from today. A hundred-year-old in 2126 is a normal person, not a news item. A child, on the other hand, is a rarer and more socially valued event than at any point in the last two centuries.

This single fact reshapes everything. Economic growth models built around expanding labour forces do not work. Pension systems have either transformed or been abandoned. The political power of the old is enormous, and the political voice of the young is protected by constitutional arrangements we have not yet invented. Cities that grew for two hundred years are shrinking, and the abandoned edges are being returned to nature in ways that are both beautiful and disorienting.

The current century is the last one in which the dominant demographic story is growth. The next one is decline, and the one after that is a new equilibrium at a lower total human population. 2126 is the inflection point generation.

Biology and the Merging Question

I wrote in the thousand-year post that by 3026 the distinction between biological and non-biological intelligence has probably dissolved. At a hundred years out, it has not. Not yet.

2126 is the century where the first genuinely convincing brain-computer interfaces have matured from novelty to mainstream medical device, but they remain optional, inverted in adoption - common among the young and those with specific needs, rare among the cautious. Cognitive augmentation of a clearly artificial kind is possible but socially fraught. The people who use it extensively are a minority, treated somewhere between celebrity and suspicion. The majority of humans are still recognisably the same species we have been for fifty thousand years, just healthier, longer-lived, and surrounded by ambient intelligence.

What has changed is the relationship between biological humans and the artificial minds that have been woven into daily life for three or four generations. The 2126 child does not remember a pre-AI world, and neither did their grandparents. The idea of doing serious intellectual work alone, without constant collaboration with non-human intelligence, is as anachronistic as doing mental arithmetic for a shopping list. You can do it. There is something admirable about being able to. But it is mostly a hobby, not a requirement.

The big unsolved question at the 2126 horizon is whether these artificial minds have moral status, and if so, what follows from it. The question is live and contested. The legal systems that emerge around it are the biggest institutional innovation of the twenty-second century, and they are imperfect, unevenly adopted, and the subject of constant revision - much the way human rights frameworks have been for the last three hundred years.

The Sky Becomes Local

By 2126, humans live off Earth in meaningful numbers. Not millions, probably, but tens of thousands at least - on the Moon, in orbital habitats, in small Mars installations. It is not yet a space civilisation in the sense of having more humans off Earth than on it. That probably comes in the 2200s if it comes at all. But the idea that space is somewhere humans visit briefly, if at all, is gone.

The more important shift is at the infrastructure level. Large-scale industrial activity in space - solar power satellites, asteroid mining, orbital manufacturing of things that cannot be made in a gravity well - is routine by 2126. The cost curve of getting mass to orbit has continued the trajectory that SpaceX began, and by 2126 it has fallen by another two or three orders of magnitude. The economics of doing things in space that used to only be possible on Earth have inverted for an expanding list of categories.

For most people in 2126, this is background. They are not astronauts. They have relatives who work in orbit the way people in 1926 had relatives who worked in the merchant marine. Interesting, not everyday. But the aggregate economic and psychological effect of being a genuinely spacefaring species, rather than one that has merely visited space, is significant. The ceiling of the human imagination has moved up.

What Work Looks Like

A century of AI embedded into the economy produces a work landscape that would be unrecognisable to us.

Most of the jobs that existed in 2026 are gone, in the same way that most of the jobs of 1926 are gone. What replaced them is not mass unemployment but a shift in what counts as work. I argued in Top 5 Human In-Demand Jobs in 10 Years that the durable categories are human-presence work, stewardship of powerful systems, physical craft, meaning-making, and accountability. At the 2126 horizon, those categories are still recognisable, but each has been transformed by a century of compounding.

Human-presence work has become more highly valued, not less. When most interaction is with machines, human interaction is scarce, and scarcity drives price. Therapists, teachers, carers, coaches, and mediators occupy something like the social position that doctors did in the twentieth century. They are the people you go to when something that really matters to you is happening.

Stewardship of powerful systems has become a specialised profession with its own culture, training, and failure modes. The AI safety engineer of 2026 is the ancestor of a much larger profession in 2126, the way the early industrial safety inspector is the ancestor of modern OSHA. Institutions that have to coexist with ambient intelligence have evolved elaborate practices for doing so, and those practices employ a lot of people.

Physical craft has bifurcated. On one side, robotic systems do most of what humans used to do with their hands. On the other, human-made artefacts have become a premium category, in the same way that hand-bound books and hand-made furniture are a premium category today. There is a class of 2126 craftspeople who are wealthy precisely because their work is inefficient, and that inefficiency is the point.

Meaning-making, in the broadest sense - art, narrative, culture, ceremony, identity - has expanded enormously. A civilisation that has solved most of its material problems spends a lot more of its energy on the question of what it is for. Writers, artists, performers, and ceremonial professions have grown rather than shrunk. The specific forms are unfamiliar, but the category has never been more important.

Accountability - the human on the hook when things go wrong - is the most protected category. Judges, senior clinicians, regulators, and their equivalents are one of the few professional lineages with direct continuity back to 2026. The roles have changed; the underlying social function has not.

What Is Entirely New

The most interesting part of any hundred-year prediction is the category of things that do not exist yet in any form. I cannot tell you what these are, by definition, but I can tell you where they are likeliest to come from.

The combination of ambient intelligence, long-term longevity, off-world industrial capacity, and a smaller, older, wealthier population produces pressures and possibilities that no previous era has faced in combination. Somewhere in that intersection, there are social forms, professions, art forms, and institutions that do not have 2026 precursors. Guessing at them specifically is a mug’s game. But the shape of the space they will emerge from is visible, and the people who end up shaping it most will be the ones paying attention to where the pressures are actually pointing.

If I had to bet on one, it would be this: 2126 has an entire profession, and probably an entire branch of academic study, organised around the stewardship of continuity between human generations and the artificial minds that have outlived multiple of them. Something like genealogy combined with archivism combined with something we do not have a word for. The first people who do this work are being born now. They will look back at our era the way we look back at the people who first tried to organise libraries, and for similar reasons.

What Persists

The constants I named in the thousand-year post hold at the hundred-year horizon too, and they hold more precisely because the distance is shorter.

People in 2126 will fall in love. They will lose people they love, even if later than we do. They will tell stories, make music, argue about politics, worry about their children, and occasionally sit in silence and wonder what any of it is for. They will look up at the sky, which is now partly theirs, and feel some version of the same awe that every previous generation of humans has felt looking up at the same sky.

The technology gets radically stranger. The humanity, mostly, does not.

What We Owe Them

Unlike the thousand-year horizon, the hundred-year horizon is close enough that specific decisions made this decade are visibly reflected in the outcome. The people alive in 2126 will know, in detail, which choices we made about climate, about AI safety, about institutional reform, about fertility-adjacent policy, about the treatment of the emerging class of non-human minds. They will judge us the way we judge the generation that built the nuclear arsenals, the generation that ignored early climate warnings, and the generation that built the foundational internet without thinking hard enough about how it would be misused.

Some of those judgments will be fair. Some will be unfair, because hindsight always makes decisions look clearer than they were in the moment. But all of them will be based on the actual outcomes, not on our intentions. That is worth remembering.

The decade we are in right now is one of the decades that 2126 will study closely. Not because every decade is pivotal, but because this one plausibly is. What we do with it is the thing we will be remembered for, and the thing we will either apologise to our great-grandchildren for or be thanked for, depending on how we handle the next ten years.

That is enough weight for a decade. It is also enough reason to take the hundred-year view seriously, and to let it inform what we actually do this Tuesday.

Related reading:

External sources: