The Big Drift: An AI Scenario

In the coming years, artificial intelligence will not simply change how we work. It will likely erode work itself as the central organizing principle of modern society.

This is not a claim about mass unemployment or human obsolescence. It is about what happens when work gradually ceases to perform the quiet functions it has long served: occupying large populations, structuring daily life, anchoring identity, and stabilizing social and political institutions. As those functions weaken, the consequences go beyond the labor market. They spread outward, into family formation, social cohesion, political legitimacy, and how people experience purpose itself. The earliest signs of this shift are already visible.

What follows is a parable, a scenario meant to illuminate a trajectory rather than forecast a timeline. Different industries, regions, and roles will experience the phases of this shift at different speeds and in different sequences. Some may skip phases entirely; others may linger in one phase for years. The purpose here is not to map the future with precision but to render visible a pattern of erosion that has already begun, and to follow its logic toward a question we will eventually all have to answer.

Consider a college graduate in 2027. She did everything right: a solid school, strong grades, reputable internships. She sends out hundreds of applications. She gets a handful of interviews but no offers. She moves back in with her parents. She watches her older brother, who graduated several years earlier, settle into a career she cannot access. The job he entered no longer exists, or exists in a form that no longer requires someone like her.

Already today, inside many organizations artificial intelligence isn’t a novelty anymore. It is embedded in everyday tools: drafting documents, summarizing information, analyzing data, supporting customer service, and assisting legal, financial, technical, and administrative work. Humans remain formally in charge, expecting that productivity will rise. But organizations are beginning to plan AI into their operations, and once a system proves reliable enough, managers stop asking whether to use it and begin assuming it will always be available.

This shift marks an inflection point because of a quiet change in hiring behavior. Entry-level roles have traditionally served two purposes: absorbing routine work and training people for more complex responsibilities. As AI systems increasingly take on that routine work, companies find they can operate with fewer junior hires. Intake narrows. Positions that would once have been backfilled simply disappear.

In early 2024, the financial technology company Klarna announced that its AI assistant was doing the work of roughly 700 customer service agents. By mid-2025, the company acknowledged the limitations of an AI-only model and began rehiring human agents to improve quality. The signal is not that AI succeeds or fails at replacing human workers — it is that firms are actively experimenting with how few people they need.

In Phase I, nothing appears broken. Operations continue smoothly. Org charts still exist. But fewer doors open.

By the late 2020s, AI agents begin operating at scale — systems capable of planning, coordinating, and executing multi-step tasks with limited human oversight. Rather than assisting individual workers, they manage workflows, integrate across tools, and handle entire categories of professional output.

The infrastructure for this phase is being built now. In March 2024, Cognition AI unveiled “Devin,” an AI software engineer capable of autonomously completing complex coding tasks end-to-end. The system was found to be underperforming, but by mid-2025 Goldman Sachs began using a version of Devin alongside human developers.

As these systems become more capable, organizations no longer merely incorporate AI into existing structures but begin reorganizing around it. Layoffs won’t come as a sudden shock but as recalibration. Teams are trimmed to where AI coordination can replace layers of supervision and execution.

Entry-level hiring does not rebound. Advancement slows. Fewer people are brought in and fewer promoted. Career ladders remain nominally intact, but their lower rungs are missing, and some rungs in the middle too.

Imagine a middle manager in his mid-forties, nearly two decades with the company. He started in an entry-level role, worked his way up, and now oversees a team of twelve. He knows the systems, the culture, and the unwritten rules. He is called into a meeting on a Tuesday afternoon. The conversation is brief. His team is being consolidated, reduced to a handful of people supported by AI coordination tools. His role is no longer needed.

As organizations accumulate experience operating with AI systems they trust, structural change accelerates. Workflows are redesigned end-to-end. Layoffs increase among middle managers and coordination-heavy roles that once existed to move information, oversee processes, and align teams.

A January 2026 IMF analysis found that almost 40 percent of global employment is exposed to AI, with advanced economies facing particularly high exposure. A 2023 McKinsey report estimated that generative AI could automate tasks accounting for 60 to 70 percent of employee time in many white-collar professions. These are not merely speculative claims; they are assessments by institutions whose job is to measure risk.

Reversal proves difficult, because competition makes AI adoption mandatory. Firms that slow down lose ground to those that do not.

The effects of layoffs become visible beyond individual firms. A large cohort of displaced white-collar workers enters a labor market with fewer places to go. Younger workers, already blocked from entry, grow increasingly disenchanted.

A new problem emerges: the tax base. Modern states are funded by wage labor. Income taxes, payroll taxes, and employment-linked consumption support everything from health care to education to infrastructure. As professional employment contracts, fiscal strain spreads across public systems.

In response, governments begin experimenting with income support — expanded unemployment benefits, tax credits, and early pilots of universal basic income. These measures are uneven and provisional. They stabilize households in the short term while leaving unresolved questions about funding, political sustainability, and scale.

As economic and institutional disruption continues, a parallel shift takes hold at the personal level. AI is no longer something people occasionally use. It becomes something they carry with them.

Wearable assistants — integrated into glasses, earpieces, and personal devices — become common. These systems provide continuous support: retrieving information, summarizing context, recalling names and histories, anticipating needs, and smoothing social interactions.

Meta’s Ray-Ban smart glasses, released in 2023, already offer voice-activated AI assistance. Apple’s Vision Pro and its rumored lightweight successors point toward a future of continuous ambient computing. Small AI-powered devices such as AI Pin, by the company Humane, and R1, by Rabbit, are early attempts at post-smartphone interfaces.

In meetings, relevant documents and prior decisions surface automatically. In conversations, forgotten details are quietly supplied. Outside of work, habits and preferences are tracked, friction reduced, and choices nudged.

The result is not a loss of intelligence but a change in how thinking happens. Cognitive effort is redistributed. Planning, recall, and judgment are increasingly mediated. Over time, people rely less on internal processing and more on external support. When the system is unavailable, even briefly, many feel disoriented — not incapable but less practiced.

Imagine a woman in her mid-thirties. Her universal basic income covers rent and essentials. Paid work is possible in theory and scarce in practice. She wakes up late. Her AI assistant suggests a routine, queues content, fills the hours. Days pass without sharp edges and do not differ much from one another. She is not in crisis, but she is also not flourishing. She is drifting.

By the late 2030s, work still matters for some but no longer structures life for large segments of the population. Time is more abundant. Obligations are lighter. Direction is less clear.

Finland’s 2017–2018 basic income experiment with a group of unemployed people found that recipients reported improved well-being but no significant increase in employment. A 2024–2025 pilot in Boulder, Colorado — this time mostly with employed people — showed the same pattern: monthly payments reduced food insecurity and psychological distress but had no significant effect on employment. Financial stability helps people survive. It does not tell them what to do with their lives.

At scale, this raises a question that economic policy alone cannot answer. When work no longer organizes daily life — when it no longer anchors identity or social contribution — what replaces it? This is the question toward which the coming years are moving — not suddenly, not uniformly, but persistently.

Who will answer it? Not economists: their tools measure output, not meaning. Not technologists: they built the displacement, not the replacement. Not politicians: they are still debating whether the disruption is real.

The question will fall to all of us: to families deciding how to raise children in a world without clear career paths, to communities trying to hold together without the rhythms of work, to individuals staring at open hours and wondering what they are for.

We will not answer it quickly. We may not answer it well. But we will have to live with the uncertainty while the old structures fade and the new ones have yet to take shape.

That is what it will mean to live through the drift.