Artificial intelligence is the most significant productivity revolution since the Industrial Revolution. That's not hype — it's the consensus of economists from Goldman Sachs to the IMF. The question was never whether AI will transform work. It's being transformed right now. The question is who gets to keep the gains. If history is any guide, the answer won't be workers.
Estimated boost to U.S. GDP from AI by 2030, under rapid adoption scenarios.
Source: Penn Wharton Budget Model, 2025 [A1]
Full-time jobs globally estimated to be "affected" by AI automation — equivalent to the entire U.S. workforce twice over.
Source: Goldman Sachs / IMF [A2]
Added to U.S. tech elites' personal wealth in 2025 alone from the AI boom — in a single year.
Source: Bloomberg / WebProNews, 2025 [A3]
Share of current U.S. work hours that today's AI technology could theoretically automate — more than half of all paid labor.
Source: McKinsey Global Institute, 2025 [A4]
The productivity gains from AI are real and enormous. Studies across writing, software development, customer service, law, and accounting consistently find AI cuts task-completion time by 15% to more than 50%, while also improving quality. In Q3 2025, U.S. labor productivity jumped 4.9% — the largest single-quarter gain in years — and AI adoption is a significant factor.
Here is the problem: productivity gains and worker prosperity are not the same thing. They have not been the same thing for 50 years. CEO pay rose 937% since 1978. Worker pay rose 5.7%. Both groups were "more productive." Only one group got paid for it.
If AI makes a worker 50% more productive, there are exactly two possible outcomes: the worker earns 50% more, or the company employs 33% fewer workers and the owners capture the gains. History tells us which one happens. The Friedman doctrine — that corporations owe everything to shareholders and nothing to workers — ensures it.
"Productivity is up. Wages are not. This is not a coincidence. It is a choice."— Economic Policy Institute [2]
The optimistic framing is that AI will create more jobs than it destroys. The World Economic Forum projects 170 million new jobs created by 2030 against 92 million displaced — a net gain on paper. [A5]
What this framing obscures: the 92 million people displaced are largely working-class people in clerical, administrative, and service roles. The 170 million new jobs are largely in AI development, green energy, and highly specialized fields requiring years of retraining. The same person who loses their data entry job is not going to become a machine learning engineer.
Goldman Sachs estimates 6–7% of the U.S. workforce faces direct displacement — roughly 11 million people. These are not abstract statistics. They are the people most likely to fall into poverty, lose health coverage, and be told by politicians that they simply need to "learn to code."
AI requires three things to build: massive computing power, vast datasets, and billions in capital. Exactly three kinds of entities have all three: Microsoft, Google, and Amazon. The AI economy isn't a level playing field. It's a toll road — and they own the tolls.
Microsoft, Google, Amazon, and Nvidia are capturing the majority of all money being spent on AI right now — and their shareholders are capturing the majority of the returns. The combined wealth of the top 10 U.S. tech leaders reached nearly $2.5 trillion by end of 2025, up from $1.9 trillion a year earlier. [A3]
The AI boom minted approximately 50 new billionaires in 2025 alone — mostly from AI startups focused on software-as-a-service, foundation models, and automation tools. Meanwhile, the workers those tools are automating received no equity, no severance uplift, and no share of the productivity gains. [A3]
The richest 1% of U.S. households owned 31.7% of all U.S. wealth in late 2025 — the highest share the Federal Reserve has ever recorded. This happened during the AI boom. It is not unrelated to the AI boom. [A7]
AI development benefits enormously from scale: more data, more compute, more users all make the model better. This creates "winner-take-most" dynamics where the largest players get better faster, making it increasingly impossible for smaller competitors to catch up. The result: permanent oligopoly in the world's most important technology. [A8]
The standard response to any concern about AI and jobs is to call it Luddism — to invoke the textile workers of 1811 who smashed machines, and to point out that the Industrial Revolution eventually created more wealth than it destroyed. This argument has three fatal flaws.
First: "Eventually" means decades. The workers displaced by the first Industrial Revolution lived through poverty, child labor, and misery in the interim. "It works out in the long run" is cold comfort to someone who loses their job next year. Second: the Industrial Revolution's gains were only broadly shared after workers organized, fought for, and won labor protections — the eight-hour day, the weekend, the minimum wage, the right to unionize. None of that happened automatically. Third: this time, the speed is different. AI is not replacing one category of job over 50 years. It is capable of replacing multiple categories simultaneously, faster than workers can retrain.
The question is not whether to allow AI. It's whether the gains from AI will be shared, or hoarded — and that is entirely a political question.
"We are building the most powerful productivity tool in human history. The gains will be enormous. The question of who captures them is the defining political question of the next 30 years."— MIT Technology Review, 2025 [A9]
In 1970, Milton Friedman declared that a corporation's only responsibility is to maximize profits for shareholders. For 50 years, that doctrine has justified cutting wages, busting unions, offshoring jobs, and stripping worker protections — all in the name of returns to capital.
AI is the perfect Friedman machine. It replaces the highest-cost input — human labor — with the lowest-cost input — computation. The savings do not go to customers in the form of lower prices (mostly). They do not go to workers in the form of shorter hours or higher wages. They go to shareholders in the form of higher margins and stock buybacks.
This is not a technological inevitability. It is a policy choice — the same policy choice made in 1980 when Reagan cut the top marginal rate to 28%, the same choice made in 2010 when Citizens United opened the floodgates of corporate money into politics. Every era has its mechanism for moving wealth upward. This is ours.
The gains from AI don't have to go exclusively to shareholders. This is a choice, and other choices are available:
None of these are radical. All of them require political will — and political will requires that the people most affected by AI displacement show up, organize, and demand it.