AI isn’t just coming for our jobs anymore—it’s already sitting in the cubicle next to us. For Center-US readers, let’s take a grounded look at what’s really happening in the job market, how far AI may go, and when we might realistically face a world where governments have to provide a base income because there simply aren’t enough jobs to go around.

1. Where we are right now: AI is biting into tasks, not wiping out work

Most serious studies agree on two things:

1. AI will touch a large share of jobs.
   • The IMF estimates that in advanced economies, about 60% of jobs are “exposed” to AI – meaning significant portions of what those workers do can be automated or reshaped. Roughly half of those jobs are likely to benefit from AI (higher productivity, higher wages), while the other half face real risk of reduced demand or elimination. IMF
   • Across OECD countries, about 27–28% of jobs are in occupations at high risk of automation when you consider AI plus other automation technologies. OECD+1
2. AI is changing tasks more than outright deleting entire occupations—so far.
   Early evidence from the OECD and ILO suggests that, up to now, AI has mostly reshaped what people do at work rather than triggering mass unemployment or huge shifts in wage inequality between occupations. OECD+2International Labour Organization+2

In other words: the job market is under pressure, but we’re not yet in “robots took all the jobs” territory. We’re in the hybrid era, where humans plus AI share the workload.

2. The current job market: turbulence, not collapse

Employers are already planning around AI

• The World Economic Forum’s Future of Jobs 2025 report finds that 40% of employers expect to reduce their workforce where AI can automate tasks. At the same time, technology trends (AI, data, automation) are expected to create 11 million jobs and displace 9 million by 2030 – a modest net positive, but with heavy disruption. World Economic Forum
• The same report projects a net 78 million new roles by 2030, with about 22% of current jobs undergoing structural change – essentially being redefined by technology, climate transition, and other macro trends. World Economic Forum+1

So near-term, the story is churn: some jobs shrink or vanish, new ones appear, and many are transformed.

Sector snapshots

• Back-office and routine information jobs (admin, customer support, basic analysis) are already being automated. Reports of companies using AI agents to reduce support staff or HR teams are no longer hypothetical—they’re news headlines. Business Insider+1
• Low-skilled work is vulnerable in some countries. A UK study projects up to 3 million low-skilled jobs lost by 2035 due to AI and automation, especially in trades, machine operations, and administrative work. The Guardian
• Programming and other high-skill tasks aren’t safe either. A 2025 analysis points out that around 40% of programming tasks could be automated by 2040, with growth shifting into AI-adjacent roles (prompt engineering, AI system design, governance, etc.). Forbes

The macro view

Big-picture forecasts still show AI as a productivity engine more than a pure job destroyer—at least through the 2030s:

• Goldman Sachs estimates generative AI could raise global GDP by ~7% and lift productivity growth by 1.5 percentage points over a decade. Goldman Sachs
• An updated 2025 Goldman analysis suggests that once AI is fully integrated, productivity in rich economies could be ~15% higher, with unemployment only about 0.5 percentage points above trend during the transition—if adoption is gradual and policy doesn’t lag badly. Goldman Sachs+1

So, near-term, the most likely scenario is more pressure on certain groups—younger workers, those in easily automated roles, women in clerical work, and people without access to retraining—rather than a universal shortage of jobs. National University+1

3. Could we actually run out of jobs?

This is the big question behind your prompt: does this end with a permanent job shortage that forces governments to provide a base income?

Economists are split:

• Historical pattern: Every major wave—steam, electricity, computers—destroyed jobs but eventually created more, often better, work. So the “standard model” says: AI will boost productivity, lower prices, create new industries and new demand, and we’ll keep finding new things for humans to do.
• AI as a break from history: Unlike previous tech, AI doesn’t just automate muscles; it increasingly automates thinking, pattern recognition, and even creative work. That raises the possibility that for the first time, machines may be able to do most economically valuable tasks better and cheaper than humans.

Some headline numbers:

• Goldman Sachs’ 2023 work suggested that generative AI could automate the equivalent of 300 million full-time jobs worldwide, affecting about one-quarter of work tasks in the US and Europe. Nexford University+1
• The OECD estimates that, if you consider all automation technologies, roughly a quarter to a third of jobs are at high risk in many OECD countries. OECD+1
• The McKinsey Global Institute finds that roughly 40% of American jobs could in principle be automated if firms fully redesigned workflows around current AI and robotics technologies. The Times

None of those models guarantee mass unemployment—but they describe a world where the default economic logic pushes strongly toward fewer humans per unit of output.

If governments don’t respond with aggressive retraining, industrial policy, and redistribution, then yes: you can get a structural shortage of “good” jobs, even if headline unemployment is managed by redefining what counts as “employment” (gig work, microtasks, etc.).

4. When might a base income become necessary?

This part is necessarily speculative, but we can build a reasonable timeline using today’s data and adoption curves.

2025–2035: Deep disruption, but still a human-plus-AI world

• By 2030, we’re likely to see:
  • AI integrated into most white-collar workflows.
  • A large fraction of routine tasks automated.
  • Ongoing creation of new roles (AI operations, safety, governance, integration, specialized services). World Economic Forum+1

Best estimate:
Through roughly 2035, most advanced economies will still have enough work overall, but with:

• Larger inequalities between highly skilled, AI-augmented workers and those in easily automated roles.
• Persistent pressure on entry-level, repetitive jobs and on workers without strong digital skills.
• Rising political pressure to experiment with partial basic incomes, negative income tax, or large-scale wage subsidies—but not yet a universal, permanent base income.

2035–2050: The tipping-point window

Assume:

• AI systems become far more capable across both cognitive and physical tasks (industrial robotics plus AI planning and perception).
• Adoption accelerates as older capital is replaced.
• Productivity gains compound over a couple of decades.

Under those conditions, it is plausible that between the late 2030s and mid-2040s, in advanced economies:

• A majority of routine, middle-skill jobs (admin, logistics coordination, basic accounting, legal support, customer service, basic programming, etc.) are largely automated.
• New job creation increasingly clusters in a few areas: high-end AI development, human-centric services (care, therapy, education, entertainment), and roles managing or constraining AI systems.
• The number of people who want full-time work at decent wages could exceed the number of such roles for years at a time, not just during recessions.

That’s the moment when a permanent base income stops being a fringe idea and becomes a stabilizing policy tool:

• To maintain consumer demand in an ultra-productive but labor-light economy.
• To prevent deep social fracture as “human-only” labor becomes less central to economic value creation.

My concrete estimate, based on the current direction of research and adoption:

A serious, widespread need for some form of government-backed base income in advanced economies is most plausibly in the window between ~2040 and 2050.

Not every country will move at the same pace; some may adopt it earlier through political choice, others later or never. But if AI keeps improving and being deployed at anything like today’s trajectory, mid-century is a realistic horizon for “we must decouple survival from a traditional job” debates to harden into law.

Beyond 2050: Post-work or polarized?

If AI reaches or exceeds human-level performance across most tasks, two macro-outcomes compete:

1. Post-work abundance (optimistic)
   • High productivity, very low marginal cost of goods and many services.
   • Broad base income, plus optional work in creative, scientific, or caregiving roles.
   • People orient life around meaning, community, learning, and experiences rather than career as identity.
2. Hyper-polarized technocracy (pessimistic)
   • A small elite that owns and controls AI infrastructure captures most income.
   • Large segments of the population are economically redundant, living on minimal transfers with limited political power.
   • Social unrest, surveillance, and authoritarian responses to keep the system stable.

Which future we get is less about the tech and more about politics, ownership, and policy choices made in the next 20–30 years.

5. What a base income world might look like—for real people

If we do end up with a government-supplied base income because traditional employment can’t cover everyone, here’s how it could feel on the ground.

The upside

• Security decoupled from employment.
  People are no longer one layoff away from losing healthcare, housing, and food. That’s huge.
• Freedom to choose “non-market” work.
  Caring for family, volunteering, open-source projects, art, community science, citizen space projects—things that don’t pay well (or at all) today could become central life activities.
• Lifelong learning as a norm.
  With AI tutors and abundant online resources, reskilling and exploring new domains becomes lower friction, even if it isn’t always “for a job.”

The downside risks

• Two-tier society.
  A top layer of highly paid AI owners, designers, and specialists versus a large population living on basic income with limited upward mobility.
• Erosion of meaning for some.
  In many cultures, identity is tightly tied to work. If traditional jobs vanish, people will need new anchors—community, creativity, exploration—fast, or risk depression, addiction, and social decay.
• Political fights over control and dignity.
  Who sets the base income amount? Is it enough for a good life or just bare survival? Are recipients stigmatized? Does the system become a tool of social control (“behave or lose your payment”)?

What people can start doing now

Even before any base income happens, individuals can:

• Shift from “job tasks” to “human advantages”. Lean into skills that are complementary to AI: complex social interaction, deep domain judgment, hands-on physical presence, leadership, storytelling, ethics, and system-level thinking.
• Build some degree of capital or ownership. This can mean traditional investments, but also participating in co-ops, DAOs, or other structures that could own slices of future AI infrastructure.
• Stay adaptable. The single most valuable trait in an AI-rich economy is the ability to continuously learn and pivot.

6. The bottom line

• Right now (mid-2020s): AI is rapidly automating tasks and squeezing certain job categories, but overall employment is still being held up by new roles and slow adoption in some sectors. We’re in the “turbulence” phase, not the “jobs are over” phase.
• Next decade (to ~2035): Expect rising inequality, intense churn in occupations, and political pressure to experiment with safety nets, but not yet a global shift to permanent base income.
• Mid-century (~2040–2050): If AI continues on its current improvement and deployment trajectory, it’s plausible that advanced economies will face a persistent shortage of good jobs relative to people who want them—forcing serious consideration of government-funded base income as a permanent feature of the system.
• Beyond that: Whether we end up in a post-work abundance society or a highly unequal technocracy will depend less on what AI can do and more on who owns it and how we choose to govern it.

For space-and-science-minded readers, AI is beginning to look a lot like a new kind of “economic gravity”: invisible, powerful, and shaping orbits of human life in ways we’re only starting to map. The challenge of the next quarter-century is to make sure that when gravity shifts, people don’t get flung into the void.