The Worst of Times

Success in creating AI would be the biggest event in human history. Unfortunately, it might also be the last, unless we learn how to avoid the risks.

— Stephen Hawking

“Heart surgery.”

“Safe procedure.”

“Only a 2% chance of mortality.”

“Incision into the groin, up through the artery, into the heart.”

My dad lays snoring in recovery. Heart rate stable. “It went well.”

It went well. What if it hadn’t?

I have not wanted to write this chapter. For some reason now, with hospital beeps around, feels like a good time.

The most quoted risk in AI is extinction, or the AI overlord scenario. That is not the topic of this book, and it is not my expertise. But there is another risk that has been a constant throughout history. It is the risk of humans. This time, empowered by AI. Long before AI becomes powerful enough to destroy us, it will be powerful enough to take our jobs. A wrecked economy arrives first, and with it, anger. Whether they wield AI or pitchforks, mobs of angry people have always shaped history. It does not take science fiction to imagine the dystopias of the past becoming our future.

A Decrepit Town

The factory towns boomed for a generation. And then their doors were shuttered, their windows broken like ripples. We can see the future by looking at what has happened.

The grandson of a man who built things now sits on the porch his grandfather built, and there is nothing for him to do in this town, or near it. Drugs. Drink. Eat. Cause trouble. When the work leaves, this is what fills the hours. The only food he can afford is the greasy, steroid-injected beef of obese cows, their arteries corroding like the rusted railways that used to connect these communities. Down long winding tunnels, lives await the past, never letting go of it. In blighted urban decay, homes lay vacant, ready for the arson that teens call weekend entertainment. Schools lay fallow, like the young minds that they once educated. Drugs fill those minds instead. Weeds, like the ones in the factory lot, for the fortunate. Fentanyl for the less fortunate. Their ravaged brains and bodies show up everywhere. In the towns. In the cities. In the streets of San Francisco. And we look away, and tell ourselves that those are them. Other people. Somewhere far away. With different life choices.

The distance is a story we tell ourselves. My friend just told me that her fifteen year old daughter asked her, "mommy, will I be homeless when I am 25?"

What do you tell that child? "It will be fine, dear." Will it? What if it isn’t?

The Irrelevant Majority

It is the future. It’s run by agents and robots. But whose agents and robots are these?

The answer is clear. They belong to whoever could afford to buy them, to whoever understood how to deploy them, to whoever had the technical literacy to make them productive. Which is to say: they belong to a very small number of people.

Construction robots print buildings, but the buildings belong to whoever owns the robots. Agricultural robots tend vast farms, but the food they produce enriches whoever controls the fleet. Manufacturing robots produce goods cheaply, but the profits flow to the few who know how to coordinate the systems.

The fractal nature of problems exists. The same inexhaustible cascade of challenges and opportunities. But only a tiny elite has the literacy to work on them. Only they can break down problems, coordinate AI systems, optimize robotic processes, and bridge domains.

For everyone else, the message is clear: you are not needed.

The construction worker who used to pour concrete isn't optimizing actuators on robotic arms. He's unemployed. The knowledge required to do that work was never built into him. He and his children can't coordinate the systems, can't speak the language of the tools, and can't contribute to that cascade of innovation. His labor is worthless now. Not because he's lazy or unintelligent, but because in a world of robotic construction, human muscle has no market value. He never learned the literacy that would let him contribute in other ways.

The pattern repeats everywhere. The agricultural laborer can't work with precision farming. She never learned how. The factory worker can't optimize manufacturing systems. The concepts are foreign. The administrative assistant can't coordinate automated workflows. The architecture is confusing. Millions of people, billions of people, like this, willing to work, wanting to contribute, desperate to earn. But they have nothing to offer that a machine can't do better and cheaper. They weren't trained for the world that emerged.

So what becomes of them?

The Dangerous Myth of UBI

Everyone seems to be clear: As Dario Amodei put it, “the key challenge in such a world won’t be incentivizing growth, but finding a way for everyone to share in the benefits.” Satya Nadella of Microsoft described, “if all the value is accrued by only a few models, the political economy will simply not tolerate it. There is no societal permission for an AI future that hollows out entire industries.” When I hear people in Silicon Valley talk about a future of such unemployment, they quickly go to the solution: Universal Basic Income.

This will solve the problem. We’ll live in such an era of abundance that with UBI and boundless intelligence and labor, the biggest problem will be boredom. People need meaning. Meaning will be the biggest problem in a world of abundance. So they say.

Here's the math that destroys the fantasy of Universal Basic Income as a solution:

A truck driver makes $50,000 to $75,000 a year. If Elon Musk sells self-driving capabilities in his trucks that cost $10,000 a year, with 50% margins, he makes $5,000 in profit per truck. Now let's be wildly aggressive and say we tax 100% of those profits to redistribute it to the displaced driver. We have $5,000 to give to someone who used to make $50,000.

The whole point of technological advancement is that it is deflationary. What used to cost a lot now costs a lot less. Taxing that fraction can never cover the costs of the original.

Scale this across the entire economy and the impossibility becomes obvious. When robots build houses, when AI handles legal work, when automation replaces millions of jobs, the productivity gains flow primarily to whoever owns the systems, in massive cost savings. What remains is a gap between what people used to earn through labor and what we could possibly redistribute through taxation. It is a chasm.

The truck driver doesn't need $5,000 in UBI. He and his kids need to be trained to be the people optimizing the logistics algorithms that coordinate those self-driving trucks. Or developing computer vision systems that let them navigate complex environments. Or managing human oversight systems that handle edge cases. They need modern AI literacy to remain economically relevant. To earn. To participate in the engine of productive growth.

The Best of Times isn't about redistributing wealth. It's about distributing the ability to generate wealth. It's about ensuring that when the robots come, when AI transforms every industry, when all of the world’s problems are endless opportunities for people to build powerful solutions, everyone has the skills to participate in building them.

What happens when we fail at that?

The Concentration

Those who do have the literacy, those who went to the right schools, grew up with the right exposure, had parents who understood what was coming, they're capturing everything.

A small group of engineers optimizes the construction robots. They don't just earn salaries. They earn ownership. They capture the massive productivity gains. They buy the buildings their robots print. They reinvest in more robots, more systems, more automation. The materials scientist who develops better solar compounds doesn't just get a paycheck. She gets equity. She captures a piece of every solar panel that uses her work. As the systems scale, her wealth compounds. The software developer who builds the coordination layer for agricultural robots doesn't just earn a wage. He takes a cut of every farm that adopts the system. As precision agriculture spreads, this income becomes generational wealth.

The returns to literacy compound. The literate few use AI to become more productive, which gives them capital to invest in and capture more of the productivity gains, which gives them resources to develop better tools, which makes them even more productive.

Meanwhile, the returns to menial mental and physical labor approach zero.

We have seen this story before.

In 1750, India accounted for almost a quarter of the planet’s manufacturing output. Britain produced barely 2 percent. The UK invested in literacy. The Industrial Revolution followed. The weaving looms began powering British economic growth, aided by the British smashing of Indian looms and, apocryphally, the thumbs of its weavers. A century and a half later, the figures had flipped. Britain made a quarter of the world's goods, and India made 2 percent. Literacy begat industry. Wealth followed industry. Power followed wealth. And power begat more power as they used this might to grow their reach across the entire planet. And those who didn’t have the literacy didn’t have the power. Poverty ensued for all of them.

What was done to a subcontinent by looms can be done to a civilization by AI, and this time the people left behind will not be on the other side of the world. They will be our neighbors. The gap doesn't just widen. It becomes unbridgeable. In a world where AI and robotics do most of the work, the only scarce input is the human intelligence that directs the systems. And if you don't have the literacy to contribute your intelligence, you have nothing of worth.

Few accrue most. Most have nothing.

A Totally Forked Future

I wake up, sunlight streaming in, waves lapping against the beach. The blue sky is the same gorgeous turquoise it has always been on every such beautiful day since the beginning of time. The birds sing their sonatas. To them, there isn't a care in the world.

But little else of the world that we grew up in remains. Many of the buildings survived, yes, but the institutions and norms eroded. The Great Reckoning, the purging, had all passed. We were left to rebuild the world anew.

Housing didn't get cheaper here. It got more expensive. Because the people who own the construction robots realized they could maximize profit by building luxury towers for each other, not affordable housing for the masses. And slum landlords for everyone else. The market always seems to tip to those who have money. Fewer people have money now.

Breakfast is anxiety. Food is technically abundant. The robots grow it efficiently. But parents can't afford much of it. The productivity gains from automated agriculture flowed to the farm owners and the robotics companies, not to lower prices. Why would they lower prices when the wealthy will pay premiums for organic, fresh, perfect produce? Everyone else gets more hormones, more steroids, more high-margin fake meat. Cheap calories for all.

The children stay home. “School” still exists, but the good ones, the ones with AI tutoring systems and adaptive learning, cost money. Those are for the children of the elite. The public schools use the old model: rows of desks, overwhelmed teachers, kids falling behind. The brilliant daughter won't become an engineer. She'll be lucky to find any work at all.

The parent searches for jobs, but there aren't many. The robots do most of what used to employ people. Some of their neighbors found work. In the service economy, catering to the wealthy elite. They counted themselves blessed. Some got gig economy jobs, doing the few tasks not yet worth automating. Income is unpredictable. Dignity is absent.

In the gleaming towers, life is spectacular. The elite live in climate-controlled environments with art and space and beauty. They travel freely. They access medical care that borders on miraculous. AI diagnostics catching diseases early, robotic surgery with near-perfect outcomes, personalized medicine tailored to their genetics. Their children attend schools where every student has an AI tutor adapting to their learning style. Alpha all the way. They grow up fluent with technology, learning to coordinate systems, break down problems, and work across domains. By the time they're teenagers, they're contributing to real research, real companies, real innovation. Their future is secured before they graduate.

In The Time Machine, H.G. Wells’ traveler zoomed into the far future, witnessing the two versions of humanity that had evolved. The Upper-world men lived lives of ease in gardens of comfort. They had drifted to be soft, beautiful, childlike and useless. Their perfect lives were enabled by the underworld people, the Morlocks. They lived underground and had adapted to the darkness, labor and deprivation. They became cruel, beastly, and inhuman. Wells was commenting on the inequalities that he saw in the England of the 1800s. He might as well be writing the same story today. With two differences. The depraved Morlocks of our future have no purpose. Not even as raw muscle. And the upper-world elite class does not grow soft. It ascends, fused with its machines, racing so far ahead that the people in the dirt below cannot even see where it has gone.

These things will take longer than we imagine. That’s part of what makes them so insidious. The decline of America’s middle class took two generations. Frogs boil slowly.

But, eventually, it breaks. Such structures of inequality in the real world rarely last long.

Angry people revolt, whether that is peacefully through a vote or by taking to the streets. Power structures shift. The only hope for the people in the gleaming towers is to suppress the underclass. They do, until eventually the pressure becomes too great and the system pops, as it always has. The masses, the so-called 99%, topple the 1%. They occupy Wall Street. They raid it. The people in their gleaming towers no longer sit enjoying the pretty view. They are the objects of witch hunts. The power structure of society inverts. Those who didn't have power are now the ones in power. When power simply flips, when the angry replace the elite, society doesn't get better. It usually trades one ruling class for another. This time with people whose primary skill is in tearing down systems. Those people are terrible at building. When they don't know how to build, they retain power by oppressing.

This was the French Revolution at 20% unemployment. The Bolshevik Revolution was born with women in Petrograd standing in bread lines, pouring into the streets demanding "Peace, Land, and Bread." The Chinese Revolution was born in villages: 85% of China were peasants, and Mao promised them the land underneath their feet.

When this happened, we saw societies traverse one of two paths.

In the places that valued freedom, it took the shape of lawlessness. Unwilling to contain freedom, an unruly form of anarchy became an uncontainable cancer. Broken states ensued. Haiti. The Congo. Yemen. Much of Mexico and Central America. But it was everywhere. Bitcoin kidnappings became so commonplace that the news didn't report them. The era of flash pandemics was upon us. Bad actors with CRISPR releasing micro pandemics to trade the prediction markets. Every year there was a new cocktail of vaccines.

It turns out that it wasn't the AI that we needed to be scared of. It was the humans. Especially humans in a world where the norms that had prevented us from pulling out a gun, or lighting a match, had faltered. These same humans were enabled by AI, and all of the multitude of tools that it unlocked. We never imagined how treacherous that would be. The countries where guns had proliferated became vigilante states. People showed their rage at the system with regular sporadic acts of violence. School shootings. Arson. Trucks driven into crowds. The regularity of these acts became part of the murmur of day-to-day life. Everyone survived, psychologically, by just praying that this all would stay in the distance, affecting someone else, not them. Once states broke, they stayed broken. In the land of the lawless, the law of the jungle won. The strongman won. Violence won. It was like gangs and mobsters fighting. That cancer spread and never recovered.

The second path was the opposite of anarchy. These were the countries that saw that order was far preferable to that chaos. It was what people called for when they feared anarchy. They called for someone, anyone, to bring back Greatness. The result: totalitarian authority. These countries all landed in the same place, whether they were led by leaders who aspired to build great nations or by leaders who saw it as their way to grip onto power. The 1920s and 1930s saw democracy replaced by authoritarianism across much of Europe: Italy in 1922, Germany in 1933, Spain in 1936, plus Poland, Hungary, Romania, and the Baltic states. Each followed the desires of frightened populations across an entire continent that turned to strongmen who promised structure. Many Germans loved Hitler because they decided that freedom was a luxury they could no longer afford. They traded it for order.

Here the danger was not that the system broke. It was that it never could.

Fighting such a regime, once you are under its thumb, is impossible. Kill a German soldier and twenty women and children would be murdered on the same spot. That was policy. And yet in history every regime like it shared a single flaw. It had to rely on people to enforce it. People tire, defect, and look away. That was the crack in every tyranny in history. AI seals it. Imagine if Hitler had an LLM. As Ilya Sutskever, a co-founder of OpenAI, warned, AI has made possible "infinitely stable dictatorships." In 1984, the ominous telescreen was a two-way screen, broadcasting propaganda while it watched and listened. But Orwell's regime still needed a person on the other side of the glass. Today we carry the telescreen in our pocket. And there is no longer anyone who needs to be watching. AI is capable of observing every breath we take, every step we take, every word we say. All the time.

China had five cameras for every citizen before the Great Break began. More countries saw how effective this was in keeping order in their nation, and they followed. But the truth is that these cameras weren't even needed. Big Tech had long ago seen that the only way to survive government hostility was to give information. So the governments didn’t need cameras to see, hear, and be in the bedroom of every citizen. They were given everything. The wealthier of these nations controlled their populations with armies of robots. Bipedal, quadrupeds, flying drones, boats. All to sustain the peace. There was no resistance.

The subtler version was more frightening, because we never saw it. Sam Altman described his real fear: an AI that reads everything you've ever written, then at exactly the right moment sends you one message, built for you, that quietly changes how you see the world. A boot on the neck can be resisted. A whisper in the ear, perfectly timed, perfectly aimed, cannot even be noticed. As in The Truman Show, everything is perfect. As in Brave New World, we drank our digital soma, numbed. If our desires are shaped to want what The State wants, is it really control?

There was a fattened class of the new powerful, a political class, safe behind big gated walls. Those who climbed the ranks were those who were best at navigating the political game. These were the ones who bore the fruits of this new world. Everyone else had no path up. Like all such political systems in history, it didn't matter if they were hard-working or educated. In fact, the system stole and punished such qualities. They were a threat to the political class.

A stable new reality. And the world continued this way for generations.

What are the chances of this future? Perhaps a 20% chance? Perhaps 2%?

What odds are slim enough to not care?

There is a spectrum, from thriving to atrophied, from utopia to dystopia. It runs through every individual and every country. Many people already live in the realities I have described. The question is not whether this future exists. It exists today. The question is how far it spreads, and to how many of us.

Relief

My dad is stable. We made it through the crisis. I sit here journaling the day after the procedure. All is well. I get to live as if nothing happened.

In the bright version of the future, we get to live on like that. My reason for writing this is not to scare you. Although, maybe it is. Sometimes it takes a rude shove to push us to act.

The real risk for my dad's procedure wasn't the procedure itself. It was what would have happened if we hadn’t taken action. One surgeon told me that he would have had a 70-80% chance of dying within a year. We have him totally healthy today because a doctor identified a problem, and then acted. I write this because what we do determines how things turn out.