In Ontario right now, a single adult on social assistance receives $733 a month. The average cost to rent a single room in Ontario is $756. The math is not complicated. The safety net pays less than a room costs. The result is exactly what you would expect.

As of July 2025, more than 30,000 people receiving Ontario Works or Ontario Disability Support Program benefits were experiencing homelessness, a 72 per cent increase since 2019. Among those who have relied on Ontario Works for more than a year, homelessness surged by 136 per cent. Seventy per cent of the unhoused recipients are single adults, the group receiving the lowest benefits. The Maytree Foundation, which obtained this data through a Freedom of Information request, called it what it is: a predictable policy failure. When they asked the Ontario government for any briefing materials exploring the relationship between social assistance rates and rising homelessness, the ministry responded that no such records exist. The government is not even studying how its own policies are creating homelessness.

Today, the Globe and Mail published new research from the University of Calgary calculating what researchers call the Homelessness Income Cut-Off, the bare minimum income needed to maintain housing after eliminating all non-essential spending. In city after city across Canada, social assistance rates fall below even this rock-bottom threshold. As researcher Ronald Kneebone put it, people on assistance are fighting like hell to hang on to their housing by using food banks and charities, and they are still losing.

Yesterday, Vividata published its Winter 2026 consumer study. Forty-nine per cent of Canadians are living paycheque to paycheque. One in three cannot pay off their credit card each month. An H&R Block survey from 2025 found that 85 per cent of Canadians feel that living paycheque to paycheque is the new norm, up from 60 per cent just one year earlier.

This is not a projection. This is not a warning about what might happen. This is Canada in March 2026. And into this economy, we are about to introduce artificial intelligence systems that will reduce the need for the jobs half these people currently hold.

In my research on AI sovereignty, on which governments control the AI systems now being integrated into public administration, there is a moment in every conversation when everything changes. It happens after the meeting, in the interregnum between discussions, after we move past infrastructure and procurement and start talking about which functions AI will actually perform. Everyone does the math and realizes that the system we are discussing is not an abstract policy question. It is a description of a future in which their role no longer exists in its current form. It is fear.

Government AI will not produce mass layoffs. It will produce mass non-replacement. A municipal AI that optimizes permitting workflows does not fire the permit officer on Tuesday, but it makes the case, in the next budget cycle, for not replacing the permit officer who retires on Friday. Attrition will do quietly what automation cannot do loudly. The fiscal logic is overwhelming, and the institutional knowledge lost will be irreversible.

There is a second displacement that no one is talking about, and it will be faster and less protected than the public sector attrition. Governments across the Western world have spent decades outsourcing work to consulting firms, not because the work required external expertise, but because hiring consultants does not increase headcount. Headcount is politically visible. Consulting contracts are buried in operating budgets. The entire model (McKinsey, Deloitte, Accenture, PwC, and the thousands of smaller firms that feed off government procurement) exists because of an accounting fiction: the government needs the work done but cannot be seen to be growing. So it pays two to five times the cost of a permanent employee to maintain the illusion of a lean public sector. In Canada, the federal government spends billions annually on professional services contracts. The ArriveCAN scandal was not an anomaly. It was a tiny, visible symptom of a systemic dependency.

AI dissolves the reason consultants were hired. A single public servant with the right AI tools can now produce the policy analysis, the scenario model, the procurement evaluation, or the program design that a six-person consulting engagement would have taken three months and $400,000 to deliver. The consulting premium, which was always a tax on the government’s unwillingness to grow headcount, evaporates. And unlike government employees, who are protected by unions, civil service rules, and the slow mechanics of attrition, consultants have no such protection. They are contractors. When the contract does not renew, they are gone. The first wave of actual job loss from government AI will not be in government. It will be in the consulting class that feeds off government. An entire tier of the professional services economy is about to discover that it was a workaround, not a workforce.​​​​​​​​​​​​​​​​

But AI is the accelerant. It is not the fire. The fire is already burning. Half the country cannot absorb a single unexpected expense. The social safety net literally pays less than the cost of a room. The systems that are supposed to prevent homelessness are producing it. AI displacement lands on top of this : not as a new crisis, but as an intensification of a crisis that is already here.

So the question everyone asks: where do the displaced workers go? is the wrong question, asked in the wrong tense. The right question is: where is everyone going?

The history of economic organization is shorter than we think, and it moves in leaps. Agrarian to industrial. Industrial to administrative. Each leap displaced the previous workforce and created a new one. Every previous transition moved workers toward cognitive work: analysis, communication, judgment, coordination. That was always the destination. AI automates the destination. The administrative economy, the one built on knowledge work, is the thing being displaced. And for the first time, there is no obvious next category of labor waiting to absorb the workforce.

Except there is. We are just not calling it what it is.

The idea is not new. In 1930, in the depths of the Great Depression, John Maynard Keynes wrote “Economic Possibilities for Our Grandchildren” and predicted that by 2030, technological progress would solve what he called “the economic problem,” the need to labor for subsistence, and that humanity’s real challenge would become how to live wisely and well with its freedom. He was right about the wealth creation. GDP grew exactly as he predicted. He was wrong about the distribution; the wealth concentrated at the top rather than spreading, so the leisure never came. Keynes saw the destination but assumed the bridge would build itself. It did not. Nearly a century later, Li Jin coined the term “passion economy” to describe platforms enabling individuals to monetize unique skills rather than commoditized labour, but her framework assumes the current system continues alongside it, a side hustle within the administrative economy rather than its successor. Yale economist Pascual Restrepo’s 2026 paper “We Won’t Be Missed” models an economy where AI drives growth and labour becomes irrelevant, but frames it as a problem of ownership rather than an opportunity for redirection. At MIT, Daron Acemoglu and David Autor are pushing a “pro-worker AI” framework that envisions humans collaborating with AI rather than being replaced, but their goal is to preserve the administrative economy with AI as a tool, not to envision what comes after it. Everyone is circling the same insight from different directions. Nobody is naming it.

A note on why this paper calls the current era the administrative economy, rather than the knowledge economy or the post-industrial economy, which are the standard academic terms. Daniel Bell and Peter Drucker named the era by its aspiration — knowledge work, post-industrial freedom, the life of the mind. But aspiration is not what happened. What happened is that every form of work, including the professions that are fundamentally about human service, was colonized by institutional administration until the administrative function consumed more time and energy than the work it was supposed to support. The U.S. Surgeon General’s Advisory on Health Worker Burnout found that nurses spend, on average, 40 per cent of every shift on documentation. A 36-hospital time and motion study published in the Permanente Journal found that patient care activities accounted for less than one-fifth of total nursing practice time – 81 minutes out of a full shift – while documentation alone consumed 35 per cent. Physicians spend roughly half their clinic day on EHR and desk work, and for every hour of direct patient care, they spend nearly two additional hours on electronic documentation and administrative tasks. According to OECD data, teachers across member countries spend half their total working time on non-teaching activities. In Canada, the proportion of working time spent actually teaching ranges from 48 per cent in Quebec to 60 per cent nationally, meaning that at best, a Canadian teacher spends four out of every ten working hours on something other than teaching. A nationally representative U.S. survey found that just 46 per cent of teachers’ time in the school building is spent teaching. The post-industrial economy did not free people from serving machines and redirect them toward serving people. It redirected them toward serving systems. The entire economy became an administrative apparatus, and the roles that were supposed to be about knowledge and human service were swallowed by process, compliance, and institutional overhead. That is why AI is not merely a labour displacement event. It is a liberation from the administrative layer that was never anyone’s actual work. The fulfilment economy is what every worker was trying to do before the administrative economy buried it.​​​​​​​​​​​​​​​​

In a recent episode of the Naval Podcast, “A Motorcycle for the Mind,” Naval Ravikant described a world in which AI gives every person the leverage to build what they want. Steve Jobs called the computer a bicycle for the mind. Naval says AI upgraded it to a motorcycle. Vibe coding (describing an application in plain English and having AI build it) means that anyone with an idea can now create a working product. English is the new programming language. The barriers between imagining something and making it exist have, for a meaningful category of creation, collapsed.

This is not theoretical. Replit, the AI-powered development platform, now has over 50 million users. Seventy-five per cent of their enterprise users are not software engineers. They are designers, product managers, operations staff, finance teams: people who had ideas but lacked the technical means to build them. Replit’s founder, Amjad Masad, learned to code in Jordanian internet cafés. His thesis was simple: if setup is the barrier, remove setup entirely. The company is now valued at $9 billion and targeting $1 billion in revenue by the end of 2026. This is not a niche product. This is the infrastructure of what comes next being built in real time.

I call what comes next “the fulfilment economy.” Not the leisure economy, which is Keynes’ framing and wrong because it implies people are not doing things, and not the passion economy, which is Li Jin’s framing and too narrow because it describes a platform feature rather than a civilisational shift. What is actually happening is a transition from an economy organized around what institutions need to an economy organized around what people want to do. Not less work. Different work. Work that serves people instead of systems.

I live in Cambodia, where the fulfilment economy already exists, though nobody calls it that. There is a coffee shop near my house that also makes noodles, that is also an evening hangout spot where people gather in the cool of the evening. There are carts that people wheel onto the street at dusk that feed entire families for a dollar. A woman sells handmade desserts from her front step. A man teaches music from his living room. Everything is multipurpose. Everyone figures it out. The economy is not organized around institutions. It is organized around people doing things for other people: feeding them, teaching them, entertaining them, gathering them together.

Here is the thing about Cambodia that the West needs to understand: the service economy is not just for the wealthy. You can get a meal for a dollar or you can get a meal for a hundred dollars. The fulfilment economy has elasticity. It serves everyone across the entire wealth spectrum because it is built around human exchange at human scale, not around institutional pricing structures designed for maximum extraction. In Canada, the opposite is true. The service economy is oriented almost entirely toward the top. Canada’s richest 1 per cent of families hold nearly as much wealth as the bottom 80 per cent combined. The top 20 per cent hold 65.5 per cent of total net worth, averaging $3.5 million per household. The bottom 40 per cent hold 3.1 per cent, averaging $82,000. If you are poor in Canada, you pay the same price for a meal, a haircut, or a yoga class as a billionaire. There is no version of the service economy that is built for you. That is not an accident. It is the architecture of an economy designed around accumulation rather than around living, and it is the reason that half the country is one paycheque from the edge. If AI is the “motorcycle for the mind,” then the Administrative Economy is a series of speed bumps and checkpoints.

It’s not that we don’t know this. It’s that we don’t *want to know it* and the things that we don’t want to know are increasing by the day: the impact of Covid, the number of people who are unhoused, Doug Ford’s cell phone records … if we don’t want know the data just disappears. It is fear that drives this; fear that makes us unwilling to confront the challenges. We see change as a threat, AI as a dystopic future and not one that is still full of possibility, even personal fulfillment. But our present is dystopic. We cannot continue to operate as we are if we want to actually be a functional society. As much as people hate AI, what it gives us is the power to understand recognize and map these trends with immediacy, at scale. As I’ve stated before pattern recognition is its highest best purpose, and we are hardly even using it. Yes it is reasonable to fear current ownership structures, but there are enough models that anyone can feed data now into a prompt window and see exactly what’s happening. And realistically – is the future that we feared not already here?

The fulfilment economy only works, only deserves the name, if it serves everyone. Not as charity. As economy. The dollar noodle cart and the hundred-dollar tasting menu are both fulfilment. The community cooking class and the private chef are both fulfilment. The difference is that an economy organized around human exchange at every price point creates work, creates community, and creates dignity at every level. An economy organized around serving only those who can afford institutional pricing creates exactly what Canada has right now: a country where the safety net pays less than a room, and half the population cannot survive a missed paycheque.

The agrarian era lasted millennia. The industrial era lasted 250 years. The administrative era is ending after roughly 70. What comes next is not another form of organized labor. It is the first economy in human history where the organizing principle is not what needs to be done but what people want to do — where the tools exist for anyone to build, create, feed, teach, gather, and care, and where the institutional machinery that used to require millions of human administrators is handled by AI.

Humanity flows like water. We adapt. The destination is not unemployment. The destination is fulfilment.

This is not wishful thinking. It is already showing up in the numbers. The global hospitality market — food, beverage, accommodation, experiences — stands at $5.5 trillion in 2025 and is projected to reach $7.5 trillion by 2030. The health and wellness market is growing from $6.8 trillion to over $10 trillion in the same period. Food services alone are expected to hit $4.7 trillion by 2030. Add these together and you are looking at roughly $23 trillion in global economic activity by the end of the decade, all of it organized around the same principle: one human doing something for another human. Cooking, caring, hosting, teaching, healing, creating experiences. These sectors are growing at 6 to 10 per cent annually while the administrative economy: the knowledge work, the processing, the institutional labour, is the sector AI is compressing. Labour is not disappearing. It is migrating. From the institutional column to the human column. From serving systems to serving people.

And this is not just an economic shift. It is a social one. An economy organized around serving each other is an economy that rebuilds the social fabric that the administrative era tore apart. The last 70 years optimized for individual achievement, individual accumulation, individual productivity measured against institutional benchmarks. The result is a society where half of Canadians report feeling lonely, where community institutions have hollowed out, where people live next door to each other for years without exchanging a name. The fulfilment economy reverses this. When your livelihood depends on feeding your neighbours, teaching in your community, creating gathering places, caring for the people around you, the social ties that the industrial and administrative economies severed begin to re-form. The economic logic and the human logic finally point in the same direction.

The question is whether anyone can survive the crossing. The crossing is happening now. The data tells us that what we are seeing is people in existential suffering, and pain and stress. And the answer is: not without a bridge.

Basic income is that bridge. Not as a progressive aspiration. Not as a response to some future AI crisis. As a response to the one we are in right now, where $733 a month is a homelessness sentence and half the country is one paycheque away from the same cliff.

The evidence is substantial. The Manitoba Mincome experiment, the Finnish basic income trial, the Stockton SEED program: they consistently show that direct cash transfers do not reduce labor participation the way opponents claim. Recipients do not stop working. They stabilize. They invest in education, caregiving, and small enterprise. The downstream savings in healthcare, criminal justice, and emergency services offset a substantial portion of the cost. The economic multiplier of money in the hands of people who spend it locally is real and measurable.

During the pandemic, Canada proved this at scale. CERB put roughly $2,000 a month into the hands of millions of Canadians and the economy did not collapse. People paid rent. People ate. People survived. And then the government took it away and called it an emergency measure, as if the emergency of being unable to afford to exist had somehow ended.

The math, when done properly, is not the obstacle. The obstacle is moral. Lawmakers still operate from the premise that income must be earned through labor, and that premise is about to collide with a reality in which there is not enough labor to go around. Mark Carney’s Liberals are on the verge of a majority: three byelections on April 13 could bring them to 172 seats. A former central banker who understands structural economic transitions has the political window, if barely, to act. Whether he spends the political capital is another question. It may be an existential one for the entire country, half of which is currently living in a state of existential financial despair.

But this is not really about Carney, or any single government. This is about a choice that every Western democracy is going to face within the next five years, and that Canada is facing right now. We have a country where the social safety net pays less than the cost of a room. Where half the population cannot survive a missed paycheque. Where AI is about to reduce the need for a significant portion of the remaining employment. And where the government’s own ministry has no records exploring how its policies create homelessness.

The question is not economic. It is not technical. It is not even political.

The question is: what kind of people are we?

Because we have already found out. We are the kind of people who will let people starve and live outside in winter. We have gone along with that. We have let it happen, but we can reverse it with proper policy, proper funding, and the kind of humanity that has been hard for us to show in the post pandemic era we can change we can treat people better. We can see the future in ourselves. The impediment to this kind of change has often been bureaucracy, as we saw in Dan Seljak’s heroic but staggering efforts to make it possible for a corner store to sell a cup of coffee. Anyone who’s ever interacted with Toronto City Council knows. Is the gradual dissolution of this kind of government a bad thing? Ask the people on the streets of Toronto.