Bill Gates is not at ease with the current pace of AI development. The Microsoft cofounder returns to his personal site (Gatesnotes) to deliver one of his starkest warnings to date.
Entitled “A Turbulent AI Era and Critical Choices to Make,” his essay marks a sharp departure from a 2023 piece in which he expressed enthusiasm for AI comparable to what he felt at the advent of the Internet and the PC.
“Even in the best-case scenario, the transition into this new AI era will be one of the most turbulent periods in human history,” he writes. “Right now, we are not preparing for it. I see no evidence that leaders, experts, and communities are rising to these challenges in a adequate way.”
The heart of the essay centers on employment. Gates contends that entry-level and mid-level jobs face the greatest risk of being simply eliminated. Sectors such as law, customer service, software development, or manufacturing could feel AI’s impact within a single decade.
For now, it is primarily white-collar office jobs that are affected, and only in what Gates calls a “modest” way. But he warns that manual labor roles will not be spared as robots become cheaper and more capable.
“Many Americans I talk to don’t realize how quickly dexterous robots are progressing,” he writes, adding that the public has been unsettled by videos of robots dancing awkwardly that have gone viral recently. He nonetheless believes that “smart” robots could begin competing with humans in certain physical tasks in construction and hospitality/food service “by the end of the decade.”
Jobs Reserved for Humans
When AI reaches a point where it can perform work flawlessly, he warns, “it will be able to operate independently, without a human needing to supervise it, and businesses will have every economic incentive to allow it.” A society with many more unemployed people will require a strengthened social safety net, he insists, recalling that factory closures in the past were accompanied by a rise in opioid-related deaths: “Imagine similar pressures, both on office workers and on manual laborers, at a national scale.”
Gates proposes an idea he calls “Human Reserved.”
As AI and robots advance, societies might choose to reserve certain jobs for humans. He cites the example of a robot delivering a diagnosis of an incurable disease: “There is no technical reason it could not do so. Yet, it shouldn’t.”
This reflection, he notes, was shaped by the care his father received for Alzheimer’s disease. According to Gates, caregivers who attended to him understood his needs even when he struggled to express them—a form of care that was “irreplaceably human” and that “no robot could have or should have provided.” He suggests domains like education or mental health could become mixed spaces, blending humans and AI, with humans remaining in the lead.
Beyond employment, Gates worries about cyberattacks on power grids, hospitals, and banks, as well as pervasive fraud and surveillance.
In terms of human relations, he cautions about the echo chamber effect AI can create, especially for children: addictive chatbots offer frictionless companionship that could dull critical thinking. “This would be the worst time for humans to lose their capacity for critical thought,” he writes. “In an era of deepfakes and misinformation that can be tailored to each individual, the ability to distinguish truth from falsehood becomes a vital life skill.”
Taxing Tokens and Robots
To finance the transition, Gates advocates taxing “tokens” and robots alike. He argues that the current tax system already pressures companies to replace workers with machines, since employing a person entails social charges whereas a robot can typically be depreciated as an investment.
According to him, these new taxes could slow the race to automation while funding a social safety net and retraining programs for those whose jobs disappear, offsetting the revenue losses from a reduced income tax base caused by unemployment.
Gates urges governments worldwide to collaborate on AI regulation, highlighting the need for US-Chinese cooperation. He compares this effort to past international coordination on nuclear weapons, aviation, or ozone protection.
He envisions a partnership among lawmakers, experts, and citizens to supervise AI. Indicating that he has already raised his concerns and proposals with leaders in the AI field, he concedes that his regulatory ideas are not yet concrete enough to win their backing.
Despite the essay’s alarmist tone, Gates stresses the positive potential of AI for health, agriculture, education, and public services, particularly in poorer countries where his foundation operates.
He says he now intends to devote most of his time to these issues, even if it means stepping back somewhat from direct involvement with the Gates Foundation.
“Waiting for people to be displaced or underemployed will be too late,” he concludes his essay. “AI represents a structural challenge to the way our economy is organized, and that requires thought and action right now.”
