AI Will Kill 33 Million Jobs—Here’s How We Replace Them All
A roadmap for full employment in the age of automation.
We aren’t doing enough in so many areas that every worker displaced by AI could be urgently needed elsewhere. New jobs could come from more teachers, elder care, mental health services, green energy retrofits, public health preparedness, climate adaptation, fact-checking and moderation, workforce upskilling, and AI oversight in the public sector. AI may kill jobs—but the wealth it generates won’t disappear. It’ll ripple through the economy, fueling new demand, new industries, and millions of jobs we haven’t built yet.
Every new technological revolution brings with it the same fear: that this time, the machines will take all the jobs. This fear is not new. It echoes through history—from the textile mills of the Industrial Revolution to the ATMs of the 1980s to the e-commerce wave of the 2000s. Capitalism did what it does best in each of these periods: it adapted. It is not always smooth or fair, but it has adapted.
Today, we stand at the edge of a new productivity wave powered by artificial intelligence. Large language models (LLMs), generative systems, and AI-enhanced automation are set to supercharge output across nearly every sector of the economy. In some industries—customer service, logistics, legal work, education, journalism, software development—AI will do the work of hundreds of thousands, if not millions, of people.
According to Goldman Sachs, AI could displace 33 million U.S. jobs over the next two decades. That’s nearly 20% of the American workforce.
This is a big number. But it’s not catastrophic—if we act with the same imagination and ambition that earlier generations showed when confronted with radical change.
Because while AI will indeed eliminate jobs, it will also unleash immense productivity gains and economic surplus. If we channel that surplus smartly— into services we've long neglected, into new infrastructure—we can not only absorb those 33 million workers, but build a more resilient society.
Let’s talk about how.
Capitalism Is a Job Creation Engine—If We Let It Be
For the last 300 years, the story of capitalism and technology has followed a pattern:
A new technology displaces some workers.
It increases productivity and lowers prices.
Consumers now have more disposable income.
New markets and demands arise.
New industries are born. New jobs are created.
The steam engine wiped out many artisanal trades—but it created factories, ports, railroads, and the middle class. The tractor hollowed out rural employment—but it fueled manufacturing and suburban growth. Computers eliminated typists and travel agents—but created entire ecosystems around IT, e-commerce, and mobile apps.
The pattern isn’t always automatic or fair. It often required public investment, labor movements, and policy interventions to ensure the benefits were widely shared. But the lesson is clear: productivity gains are not the end of work. They are the beginning of something new.
AI Will Boost Tax Revenues—Let’s Put Them to Work
AI’s potential GDP contribution is enormous. McKinsey estimates AI could add $2.6 to $4.4 trillion annually to the global economy. Even a modest slice of that—properly taxed—would create enormous fiscal capacity for governments.
Let’s say AI boosts U.S. GDP by 10% over 10 years (a conservative estimate). That’s nearly $2.5 trillion in added economic activity. We’re looking at hundreds of billions in new annual revenue.
That’s money we can invest in the kinds of jobs we want to create. Not just because they soak up displaced labor—but because they make our lives better.
Where We’re Failing Now—and Where the Jobs Are Waiting
Let’s be honest: the United States has never been a fully optimized society. We’ve never fully staffed our classrooms, nursing homes, mental health centers, or disaster response units. We’ve cut public health funding, neglected elder care, and left millions without access to basic services.
That’s not a labor shortage problem—it’s a priority problem.
We can fix that.
Here’s one simple example: The U.S. has about 3.5 million teachers. The average class size is roughly 24 students. What if we capped classes at 12 instead? You’d need another 3.5 million teachers—and the impact on education outcomes, attention, and even eventually productivity would be transformative.
That’s just one piece of the puzzle. There are dozens more. We can meet the AI jobs challenge by investing in education, elder care, mental health, green infrastructure, public health, climate adaptation, digital civics, upskilling, and AI oversight—jobs that serve people, not just machines.
These jobs are not fantasy. They are in fields that already exist. They are necessary. And in many cases, they have been starved of funding and attention for decades.
AI Also Creates Jobs—Just Not Where You Think
One of the most overlooked facts of AI transformation is that it requires massive human effort to succeed. Consider:
Training and fine-tuning large models requires labeled data, ethical oversight, and subject-matter expertise.
AI alignment requires interdisciplinary teams across philosophy, policy, and psychology.
AI deployment in healthcare, law, and education requires careful oversight, customization, and human judgment.
Workforce upskilling will be essential—every displaced worker needs pathways into AI-era industries.
The WEF projects millions of new jobs in AI safety, testing, moderation, prompt engineering, and regulatory compliance.
So while 33 million jobs may be automated away, millions more will be automated adjacent—if we train for them.
And don’t forget the multiplier effect: all the wealth generated by AI-driven productivity gains doesn’t vanish—it recirculates through the economy, creating new demand and, with it, new service jobs.
Conclusion: Absorb the Shock, Shape the Future
The AI transformation is real. It will displace work. But it will also create abundance. The key question is whether we will use that abundance to invest in society—or hoard it.
We’ve been here before. The challenge of reabsorbing 33 million workers is no more daunting than the shift from farms to factories, from manufacturing to services. We’ve done this. We can do it again.
But it won’t happen automatically.
It requires public policy, imagination, and commitment to human flourishing. If we want AI to be a net good, we have to shape the transition—not just endure it.
And if we get it right, AI won’t just replace jobs. It will liberate us to do the work that matters most.
Sean Richey is a researcher focused on political communication, emerging technologies, and the intersection of AI and society.
Subscribe to get the next post in your inbox.