Bill Gates is doing something valuable in his new Gates Notes essay and his Sept. 29 interview on The Ezra Klein Show: taking both the promise of AI and the people who may lose work because of it seriously. He sees a chance to make health care, education, and other essentials more accessible. He also asks what happens to workers and communities while that future is being built. That combination of hope and responsibility deserves support.
The most important part of the interview begins around minute 27, when Klein asks why Gates thinks AI will take jobs at significant scale. Gates does not claim that mass unemployment has already arrived. In fact, he says AI has created more jobs than it has destroyed so far, especially around data centers and model infrastructure. But he argues that the decisive question is reliability. Once an AI system is good enough to complete a whole task, the labor market does not necessarily create an equal number of new roles somewhere else.
That is the hinge. Gates says coding is the first profession where AI has crossed that threshold. He expects more thresholds soon in accounting, legal work, telesales, and customer support. The interview transcript captures the exchange: Gates carefully distinguishes the work AI may create from the jobs it may replace. Cheaper intelligence can increase demand in some fields, but there is no guarantee that every lost paycheck will be replaced.
His example is radial tires. Better tires lasted much longer, but people did not drive four times as much simply because tires improved. The result was not a new golden age for tire workers. It was fewer workers needed to produce the same useful service. Gates is applying that logic to AI: where demand is elastic, cheaper work can expand the market; where demand is not elastic, the savings flow away from wages.
For a site called Income For Everyone, this is the central point. When a person loses a job, they lose a source of income even if the work still gets done. If a company can replace a worker with software, the cost moves from a paycheck to an AI bill. Gates calls that the “token budget” instead of the human salary budget. He is giving us a clear way to see why the benefits of AI will need a new path into people’s lives.
Gates Is Right to Start This Conversation Now
Gates is right to focus on speed. Past technological transitions were hard for many workers, but they often unfolded over generations. AI already runs through the tools people use at work, and its reach can expand quickly across offices, call centers, classrooms, clinics, and software teams. Preparing early gives families, employers, and governments more room to adapt.
He is also right to look beyond the promise that new jobs will appear. They may, and he acknowledges the jobs AI has created already. But a worker whose occupation changes faster than they can retrain still needs to pay the bills. New graduates need a way into the workforce. Gates treats those practical questions as part of the AI conversation, where they belong.
Gates’s essay names three policy directions: build serious domestic and international AI institutions, reserve some work for humans, and consider taxes on AI or robots as the tax base changes. His “human reserved” idea is especially thoughtful: in care, health, education, and other deeply human settings, society may choose to keep people at the center even when machines can perform parts of the work.
That proposal matters. There are things a society should not automate just because it can. A diagnosis, a classroom relationship, end-of-life care, child care, and mental health support are not just transactions. They are part of how people experience dignity, trust, and belonging.
These are constructive starting points. They also invite the next question: how will everyone share in AI’s gains, including people whose work cannot be reserved?
Build the Income Policy Alongside the Technology
Klein presses Gates on the moral tradeoff: if AI brings job loss, social disruption, and dangerous new capabilities, why accelerate it at all? Gates’s answer is that AI could also lower the cost of health care, education, food, housing, and energy over a 20-year period. It could help farmers in India, improve medical advice in underserved languages, and bring useful tools to people who have been ignored by existing markets.
That is a real upside, and Gates also recognizes the timing problem. Job changes can arrive before lower prices for rent, care, electricity, and food. His concern gives us a reason to build income security while we work toward those benefits.
That is where Universal Basic Income belongs in the conversation. Elon Musk has said that automation may lead to a form of UBI; in later remarks, he has imagined a more expansive “universal high income.” We share the hope that technology could make such security possible. But income will not simply appear when AI becomes more capable. People must decide who qualifies, what level of support is meaningful, how it is funded, and how payments reach everyone reliably. Those choices take legislation, budgets, public trust, and time.
Gates’s proposals offer a place to begin. If AI changes where economic value is created, tax policy may need to change with it. A UBI funded in part by AI-driven gains could give everyone a dependable floor while human-reserved work, public services, and labor protections continue to matter. This is a policy addition to Gates’s agenda, not a claim that he endorsed a particular UBI design.
A serious AI transition plan could include:
- An AI dividend or compute tax that sends revenue back to the public, not only to retraining programs.
- Automatic cash support that rises when layoffs, underemployment, or labor-force exits increase in AI-exposed occupations.
- Shorter workweek experiments so productivity gains become time gains instead of only margin gains.
- Human-reserved care and education roles with public funding, strong wages, and AI used as support rather than replacement.
- Portable benefits and wage insurance for workers forced out of occupations before new work actually exists.
Each idea needs public debate, careful design, and a way to measure whether it helps. That work can begin now, while there is still time to make the transition fairer.
Why This Special Issue Matters
Gates has spent his life helping bring new technology into everyday use. His willingness to talk plainly about job loss, and to offer ideas such as human-reserved work and AI taxation, makes this an unusually useful contribution. He is inviting a serious discussion about who benefits from AI and how to protect people through the transition.
We should take that invitation seriously. Musk’s vision of income security in an automated future is hopeful, but a prediction is not a payment system. A workable UBI needs public policy before large-scale job disruption leaves families waiting for help.
AI may create new work and make some goods and services cheaper. If Gates is right that many tasks will move from payroll to compute, UBI can help make the resulting prosperity broadly shared. Designing that system now is a practical way to honor both his optimism and his warning.
The lesson from Gates’s essay and the Klein interview is an encouraging one: the outcome is still ours to shape. We can support the breakthroughs he sees and prepare policies that put a dependable share of their gains in people’s hands.