1956 · The switchboard: A word is coined for machines that might think.
Rosa is nineteen and connects calls by hand: plug, ring, listen, unplug. It is one of the most common jobs a young woman can get. That summer, a handful of researchers at Dartmouth propose a project with a new name.
1985 · The machine floor: Robots arrive on the factory floor.
Rosa’s switchboard is gone. She retrained as a bookkeeper at the plant across town, where the first robot arms now weld car bodies. The night shift gets shorter every year.
1997 · Checkmate: A computer beats the world chess champion.
Rosa’s granddaughter Maya is three. The highlight moves to her now: from here on, we follow Maya. IBM’s Deep Blue defeats Garry Kasparov, and for a while the world treats it as a curiosity.
2012 · Machines learn to see: Deep learning breaks through.
Maya starts college on a scholarship. In a Toronto lab, a neural network called AlexNet crushes the ImageNet benchmark, and the tech industry pivots almost overnight.
2023 · The fastest app in history: A chatbot reaches 100 million people in two months.
Maya, 29, answers support tickets for a software company on a team of forty. Then the team gets an AI assistant that drafts replies. New hires suddenly perform like veterans, and Maya starts to wonder why the company needs so many of them.
2024 · Exposure: Capital pours in. The team shrinks.
Forty agents become twelve. Maya keeps her job by moving to escalations, the angry or unusual cases the AI can’t close. Several friends from the old team are still looking.
2026 · Today: The present, as of the day you’re reading this.
Maya today: she supervises the AI agents that now answer the tickets she used to.
2027 · The agents clock in: Software stops waiting to be asked.
AI agents no longer just draft. They book, refund, fix and file whole tasks on their own. Maya, 33, leads four people and forty agents. Her job is writing the rules the agents follow and catching the cases they get wrong. The company has stopped hiring juniors.
2028 · Machines on the street: The machines leave the screen.
Driverless taxis run all night, delivery robots share the pavement, and the warehouse by the river has more machines than people. Maya’s neighbour Sam drove a cab for twelve years. He retrains to service the fleet that replaced him. Maya’s job holds, but she rarely does anything by hand any more.
2029 · Power and pay: Intelligence turns out to run on electricity.
The data campus by the river now uses more electricity than the rest of the district. The profits flow to a few firms; the noise, the land and the price spikes land on everyone else. Maya, 35, is promoted. Her younger cousin, with a new degree, can’t find a first job.
2030 · The vote: The city votes on who gets the gains.
Nobody argues any more about whether AI is changing work. The argument is about who gets what it makes. The city holds a referendum on an automation levy with three answers on the ballot: leave it to the market, tax it and share it, or make the basics free for everyone. Maya, 36, volunteers at a polling station.
2035 · The fork: Same technology. Three different societies.
Five years after the vote, the city has taken one of three roads. It came down to who owns the machines and how the gains are shared. Choose the result: the city, and Maya’s life, will follow it. The evidence below is what we know from real cash-transfer experiments.
2050 · Horizon: A city that shares the dividend.
Maya is 56. An automation dividend arrives every month, funded by a levy on AI profits. She works four days coordinating home care for older neighbours, a job no model does well, and spent two paid years retraining for it.