Discussions about AI and employment now dominate the news cycle. Every week brings a fresh round of "X percent of jobs at risk" headlines, fresh layoff announcements at companies that just announced record profits, and a fresh round of think-pieces about retraining.
Most of the conversation assumes the goal is to keep humans employed at jobs that look roughly like the jobs we have now. That assumption is the part that should be questioned.
I am not going to spend much time on the Calvinist obsession that says "you must work" — even if the work is busywork, meaningless to the worker and to society. It is a stupid belief that has been quietly contributing to the destruction of the biosphere for at least a century, and it is the unexamined background assumption sitting under most of the AI-jobs panic.
We have enormous brainpower. We could figure out a much more logical world. We are not going to iron that out in this essay, but it is worth naming the assumption before we start.
The more interesting question is buried under the panic: what could the freed-up human time actually be redirected toward?
This essay was first written in 2024, when AI-induced job displacement was still mostly hypothetical. Two years later, the displacement is measurable:
The full data and historical comparisons are at scovert.com/ai-vs-jobs.html. The short version: the displacement is real, it is concentrated, and it disproportionately hits people who never got the chance to develop deep judgment in the field before it was automated.
Those freed-up hours have to go somewhere - whether it's logically planned or not. The question is whether the somewhere is productive, or whether it is despair, isolation, and political destabilization.
The hardest part of the post-labour conversation is not technical, economic, or even political. It is religious. The North American work ethic was forged in a Calvinist theology that treated employment as a sign of moral worth and idleness as a sign of poor character. That belief has been secularized for so long that most people don't recognize it as a belief at all — it just feels like common sense.
It is not common sense. It is a four-hundred-year-old artifact of a particular religious tradition, weaponized over time into a story that helped industrialists keep wages low and labor cheap. The cost of that belief now is an entire society that confuses "having a job" with "contributing to anything."
Unbundle those two things and the AI-jobs panic looks completely different. The question stops being "how do we keep people in jobs that don't need them?" and becomes "how do we direct freed-up human capacity at things that actually need doing?"
The term is academic. The idea is not. Post-labour economics asks what happens when the link between paid employment and survival is broken — either because automation has eliminated most of the wage-earning jobs, or because the jobs that remain don't pay enough to live on, or both.
It used to be a fringe topic. As of 2026, it is one of the fastest-growing areas of economic and policy interest. The reasons are obvious: AI in the writing/design/coding stack, robotics in manufacturing and logistics, agentic systems in customer service and administration. None of those displacements are theoretical anymore.
The serious people working on this aren't proposing utopia. They are asking practical questions like: if a 35-year-old paralegal whose job just got automated has 40 years of working life left, what are the options? "Retrain for a job that will also be automated in five years" is not a satisfying answer.
"Universal Basic Income" is one option being piloted in multiple jurisdictions. "Job guarantee in environmentally and socially useful work" is another. The choice between them is, in part, a question about what kind of society we want.
The work below has been needed for decades. None of it has been done at the necessary scale, mostly because there has never been a profit motive aligned with doing it. AI freeing up human time changes that math — if the political will exists to pay people to do it. That's the real question. Not whether the work is needed.
Planting trees on deforested or degraded land. Trillions of tree-planting hours have been talked about; only a fraction have been actually funded.
Rebuilding wetlands and marshes to improve water quality, restore wildlife habitat, and mitigate flooding for the towns downstream.
Rebuilding dunes, mangrove forests, and coral reefs to protect shorelines from erosion and storm damage that will only get worse.
Terracing, re-vegetation, and erosion control to prevent agricultural soil degradation and desertification. North America loses topsoil faster than it forms it.
Removing obsolete dams, restoring natural river channels, replanting riparian vegetation. Aquatic habitat recovery has direct effects on drinking-water quality.
Creating wildlife corridors, constructing nesting sites, reintroducing native species. The biodiversity collapse has decades of restoration work behind it.
Green roofs, community gardens, urban forests. Mitigates heat-island effects, improves air quality, demonstrably reduces violent crime in adjacent neighborhoods.
Removing plastic pollution, abandoned fishing gear, and other debris from coastlines and ocean environments. The Pacific Garbage Patch is not going to clean itself.
Field surveys, ecological-indicator monitoring, baseline data collection. AI can process the data; humans have to collect it in places AI can't go.
Local environmental education, capacity building, participatory restoration. The work that can't be centralized — only locally led.
Soil remediation, wetland restoration, and re-vegetation of areas impacted by tar sands extraction. The reclamation bond money has been collected; the work has barely started.
Restoring contaminated land at abandoned or inactive mines. There are tens of thousands of these sites in North America alone, mostly orphaned by bankruptcy.
Brownfields, landfills, contaminated industrial facilities. Soil and groundwater remediation. The EPA's Superfund site backlog is decades old.
Containment, cleanup, and ecological restoration after marine and inland spills. Faster mobilization saves more shoreline and more wildlife.
Beach nourishment, dune restoration, shoreline stabilization. Defending the towns and infrastructure that are being eaten by rising sea levels.
A rough calculation. Assume Goldman Sachs is even half right about the net 16,000 jobs per month figure. Over a decade, that compounds into roughly two million displaced full-time workers in the U.S. alone — on the low end of credible estimates.
Two million workers at 2,000 hours per year is four billion hours per year of freed-up human capacity in the U.S. alone. Compounded across the OECD, the figure runs into the tens of billions of hours per year, possibly into the hundreds of billions over a decade.
What is currently happening with most of that freed time? Best case, retraining for the next AI-exposed job. Worst case, opioid epidemic territory — despair, isolation, decline. Neither is what was supposed to happen.
If even a fraction of those hours got directed at the restoration work above — tar sands cleanup, wetland rebuilding, urban greening, ecosystem monitoring — the impact would dwarf anything any government environmental program has attempted in the last fifty years.
Redirecting freed labor toward environmental restoration is not just an ethical argument. It is also the economically defensible one once the long-term costs are honestly priced.
The economist's term for this is "internalizing externalities" — pricing in costs that the market has been ignoring. Restoration work is the labor side of that math. AI displacement is the supply-of-labor side. They line up.
None of the above is automatic. The math works on paper. The politics is harder than the math.
Who pays? Restoration jobs don't generate profit the way private-sector work does. Funding has to come from public sources (tax revenue, carbon-credit markets, restoration bonds), from corporate social-license obligations (the "AI Industry Pact" idea that some of the AI productivity gains get redirected toward public-good work), or from some hybrid. Each has political enemies.
Who decides what gets restored? Tar sands cleanup is not the same political project as urban tree planting. The communities most affected by industrial pollution are usually the communities with the least political power. Any restoration program has to grapple with that.
Who manages the work? Environmental restoration at scale needs ecologists, engineers, project managers, and on-the-ground labor in roughly equal measure. The current displaced workforce is mostly desk-job-trained. Bridging that gap honestly is a real cost, and most "green jobs" rhetoric pretends it isn't.
International coordination? Soil and ocean don't respect borders. Some of the highest-leverage restoration work is in places where current geopolitics makes employment programs hard. That doesn't mean the work shouldn't happen; it means the planning has to be more sophisticated than "we'll figure it out as we go."
The point is not that the politics is easy. The point is that the labor side is solving itself whether anyone plans for it or not. The longer we pretend that "more jobs in the old industries" is the answer, the more freed labor we burn for nothing.
Most of the AI unemployment conversation accepts the frame that the goal is to keep humans doing the jobs they have now. I'd argue that that's a TERRIBLE frame of mind.
The real frame is: we have, simultaneously, a freed-up labor supply (because of AI) and a deep backlog of needed restoration work (because the biosphere has been damaged for decades and the work hasn't been profitable). The question is whether we connect those two facts deliberately, or whether we let one of them fail loudly while pretending the other doesn't exist.
This is one specific application of a broader pattern that's worth knowing about: AI is starting to take on entire categories of work that used to require many humans. At the planetary-coordination scale, that's already happening — Google Flood Hub uses AI to direct human and capital resources toward flood-vulnerable regions across 80+ countries. Singapore's national workforce platform uses AI to steer training and employment toward strategic sectors. The pattern is real, the tools exist, and the scale is global.
What's needed is the political and institutional will to apply the same pattern to restoration work specifically. The biggest piece holding that up is the unexamined Calvinist assumption that "having a job" is the only way to legitimately contribute to society. Drop that assumption and the rest gets easier to design.
Rather than succumbing to fear and anxiety about AI-induced unemployment, we have the chance to use this moment of freed-up labor to transition toward a more sustainable and resilient future. By reframing the discourse and harnessing the labor that's already coming free, both the planet and the people displaced by automation come out ahead. The alternative is to let the freed time be burned on despair, scrolling, and political destabilization. That's not a small choice.