Artificial intelligence is now cheaper than many human workers, and Bill Gates thinks that's a problem. In a sweeping 6,000-word essay, the Microsoft co-founder warns that AI is about to change everything and that the world is not ready. His solution? A new kind of tax designed to make companies think twice before replacing people with machines.

Gates's essay lands amid growing anxiety about AI's effect on jobs, privacy, and security. He argues that the disruption ahead will be unprecedented, comparing it to the arrival of the internet or the industrial revolution. But unlike those earlier shifts, he says, the pace of change may be too fast for governments to catch up. The result could be mass unemployment, social instability, and the concentration of enormous power in the hands of a few technology companies.

Key facts from Gates's AI essay

• Bill Gates released a 6,000-word essay warning that AI could cause unprecedented disruption.
• He fears AI could put weapons of mass destruction in the wrong hands.
• He also warns that the job market could be devastated, with world-altering consequences.
• Gates calls for a new category of “Human Reserved” jobs that only humans would perform.
• He proposes an AI token tax and robot tax to remove the current financial incentive to replace workers.
• The current tax system encourages automation because payroll taxes cost employers money, while AI expenses can be written off.

Why Gates is worried

It would be easy to dismiss Gates as a late convert to the AI debate. He was famously slow to grasp the potential of the internet in the mid-1990s, and he has admitted that he underestimated how quickly AI would improve. But his latest essay is not a sudden revelation. Instead, it reads as a warning from someone who has watched the technology develop from inside some of the world's most powerful companies.

The former Microsoft CEO doesn't see AI itself as the only problem. He is equally alarmed by the attitude of policymakers who seem content to let the market sort everything out. In his essay, Gates writes that the disruption will hit vulnerable workers first, widening inequality and creating a class of people who feel left behind. He warns that if governments do not act, the backlash against AI could be severe and could slow innovation even in areas where the technology would clearly help.

The “Human Reserved” jobs proposal

One of Gates's most talked-about proposals is a category of jobs that would be reserved for humans. He calls these roles “Human Reserved” positions, and he includes caretakers and teachers as examples. The idea is that certain jobs carry a human element that should not be automated, no matter how capable machines become. He argues that society should decide, in advance, which roles remain off-limits to AI.

On its face, the concept sounds thoughtful. But as Gates himself concedes, it raises difficult questions. Who would decide which jobs are reserved for humans? How would those rules be enforced across different countries and industries? And could the label become a way to limit wages or trap workers in jobs that might otherwise evolve? Critics have also noted that many caretaking and teaching roles are already underpaid and undervalued. Making them “reserved” for humans would not automatically make them good jobs.

An AI token tax

Gates's other proposal is more concrete, and more controversial. He suggests taxing AI tokens and robots, much like governments tax tobacco, carbon emissions, or payroll. The idea is to change the economics of automation by making it more expensive to replace a human worker with an AI model. If every token processed by a large language model carries a small tax, then companies that use AI to replace call-center agents, writers, or data-entry clerks would face a new cost.

Why does that matter? Under the current tax code, employers must pay payroll taxes for every human worker, including Social Security and Medicare contributions in the United States. Those taxes add up quickly. AI, on the other hand, is usually treated as a business expense, which means companies can deduct the cost of AI tools from their taxable income. That creates a built-in incentive to replace people with software, even when the quality of work is not equal.

An AI token tax could flip that equation. It would level the playing field between human labor and machine labor by adding a cost that reflects the social impact of automation. Revenue from the tax could be used to fund retraining programs, support displaced workers, and build the safety net that Gates argues will be necessary as AI disrupts entire industries.

This is not a brand-new idea. Gates has talked about taxing robots before. In a 2017 interview, he suggested that a robot that replaces a human worker should pay taxes at a rate similar to the worker's income taxes. That proposal was widely debated at the time, with some economists arguing that it would slow innovation and make countries less competitive. The AI token tax is essentially a modern version of that idea, updated for the era of chatbots and generative AI.

Supporters of the tax point out that it would force companies to account for the hidden costs of automation. When a business fires hundreds of workers and replaces them with an AI system, the public often ends up covering the costs of unemployment benefits, housing assistance, and health care. A token tax would make part of those costs visible in the price of the AI service.

The complications of taxing AI tokens

Implementing a tax on AI tokens would be far from simple. Not all tokens are the same. A token used by one of the advanced reasoning models costs far more than a token used by a basic model. The tax would have to account for different models, different pricing structures, and different use cases. Should a token used in medical diagnosis be taxed the same as a token used to generate marketing copy? Should student users get a tax exemption? And how would tax authorities keep up with the rapid pace of model releases?

There is also the problem of loopholes. Companies could route their AI workloads through developers or offshore providers to avoid taxes. They could also create their own open-source models and run them on their own hardware, making a token-level tax nearly impossible to enforce. Governments would need to define what counts as a taxable token, and they would need to update those definitions regularly. That kind of regulatory agility is rare, especially in tax law.

Another question is whether the tax would actually protect jobs. Some economists argue that a token tax would simply make AI more expensive for everyone, slowing down useful applications in health care, education, and scientific research. Others say the tax would not be large enough to matter, especially if AI continues to get cheaper. If the price of AI falls faster than the tax rises, companies will still have a strong financial reason to automate.

Killing the economic incentive to replace humans

Despite the challenges, Gates's core argument is difficult to ignore. The current system is rigged in favor of automation. Every time a company replaces a worker with an AI model, it saves on payroll taxes, benefits, and other labor costs. The government loses tax revenue, and the displaced worker loses income. In that sense, the public is already subsidizing the transition to AI, but without any of the guardrails that would make it fair.

An AI token tax would not stop the transition, and Gates is not proposing that it should. The goal is to slow down the most disruptive kinds of job replacement, and to generate revenue that can be used to help people adapt. That makes it different from an outright ban on AI, which would be impossible to enforce and would come with enormous economic costs.

Business groups are likely to fight any new tax on AI. They will argue that it puts companies at a competitive disadvantage, drives investment overseas, and penalizes the very innovation that could solve problems like climate change or disease. But Gates's essay also speaks to a broader political reality: millions of voters are anxious about AI, and they are looking for leaders who will do something about it. A tax on AI tokens could be surprisingly popular, even if the details are ugly.

There are also deeper questions about what kind of society we want to live in. If AI can do most routine cognitive work, then the value of human attention, care, and empathy may increase, not decrease. But that will only be true if we create structures that reward those qualities. The “Human Reserved” label may be too vague to work, but it points to a real need: to decide which parts of life should remain human-centered, even when machines could do the job.

AI is cheaper than workers today, and it will become even cheaper tomorrow. Bill Gates wants to change the calculation by making automation pay for the disruption it causes. His proposals may be messy, incomplete, and politically difficult, but they are an honest attempt to answer the question that politicians have been avoiding. If we are not ready for AI, the time to start getting ready is now.

Source:PCWorld News