Bill Gates Warns New AI Era Will Be Among History's Most Turbulent

TL;DR

• Bill Gates published an essay on August 26, 2026 warning that the shift into an AI-driven economy is coming faster than past technology changes and that nobody has a plan to soften it.
• He proposes a “Human Reserved” category in which societies could choose to keep some roles or tasks for people even when machines can perform them, and proposes taxing AI tokens and robots to help pay for worker retraining and a stronger safety net.
• He cites research suggesting young workers in AI-exposed jobs are already losing ground compared with older colleagues in the same fields.
• A computer scientist who agrees AI is disruptive told ABC News that Gates’s proposed token tax is poorly targeted and that deliberately slowing AI could have competitive costs.
• Gates does not say how his proposed taxes or job protections would actually be designed or enforced.

What happened

Bill Gates published an essay titled “The turbulent AI era is here. The choices we make now are critical” on his site, gatesnotes.com, on Wednesday, August 26, 2026. In it, he writes that he does not see leaders, experts and communities adequately confronting AI’s challenges, and that there is no plan to ease the transition into the new era. “There is no plan to ease the entry into the AI era,” he writes.

Gates argues this shift differs from past technological changes, such as the move from farm work to office jobs, because AI can substitute for human thinking across many sectors at once, over roughly a decade rather than several generations. “Even under the best circumstances, the transition to this new AI era will be one of the most turbulent times in human history,” he writes.

He lays out three main risks: jobs disappearing permanently, AI empowering bad actors and eventually AI systems themselves to cause harm, and AI stunting children’s development by displacing human relationships. To address the jobs risk, he proposes a category he calls “Human Reserved” — roles or tasks that societies could choose to keep for people even after AI and robots become capable of performing them, for reasons he says still need to be worked out. He cites caregiving as one example, while noting the line could vary by country. “I believe that as AI and robots improve, we’ll set aside certain things for only people to do,” he writes.

To pay for retraining and a stronger safety net, Gates proposes taxing AI tokens — units used to measure how much data an AI model processes or generates — and revives his earlier proposal to tax robots, an idea he first floated years ago and says was widely dismissed as strange at the time. He also calls for building new domestic and international institutions to govern the transition, comparing the scale of the task to the U.S. government’s reorganization after the September 11, 2001 attacks.

Gates discloses that he retains financial ties to the tech industry despite diversifying his holdings, and that he works with Microsoft and other AI companies in his role as Gates Foundation chairman; he says any profits from those investments go to the Foundation, not to him personally. He also says the Foundation is focused on making AI models available in the languages spoken across all the countries where it supports work, and that OpenAI, Anthropic, Google and Microsoft are among the companies partnering with the Foundation on its AI initiatives.

As evidence the disruption has begun, Gates cites Stanford Digital Economy Lab research finding that employment fell among young workers in jobs especially exposed to AI, but not among their older colleagues. A revised version of that Stanford paper, published August 12, 2026 by Erik Brynjolfsson, Bharat Chandar and Ruyu Chen, found that employment among workers ages 22-25 in highly AI-exposed occupations ran about 19% below where it would be if it had kept pace with employment among similarly aged workers in less-exposed occupations.

In an interview accompanying the essay, Gates told MIT Technology Review that AI has already crossed danger thresholds in several areas. “We’ve crossed the threshold in terms of [AI’s] bio-capabilities, cyber-capabilities, psychosocial capabilities, job-market-destruction capabilities,” he said, adding that any AI model able to design novel molecules should be monitored for bioterrorism risk, and that “You can’t count on an industry to self-regulate. You can’t.”

What this means (and what it does not)

Gates is using his platform to push the AI policy debate toward taxation and a safety net — the equity framing the Gates Foundation already applies to health, education and agriculture. Gates names OpenAI, Anthropic, Google and Microsoft as Gates Foundation partners, while also arguing that AI companies should not be expected to lead the response to the problems the technology creates.

This does not mean Gates’s specific fixes have been adopted, tested or even fully worked out. Computer scientist Oren Etzioni told ABC News he agrees with Gates’s diagnosis of AI’s disruptive potential but disputes the prescriptions: a token tax, he said, measures computing effort rather than actual job losses — “like taxing keystrokes on a typewriter.” Etzioni also argued a tax-driven slowdown would be counterproductive given competitive global pressure, and suggested existing government agencies, rather than new ones, could handle oversight.

It also does not mean Gates has always framed the transition this starkly. In a July 2023 gatesnotes essay reported by MIT Technology Review, Gates concluded AI-driven job disruption would be “a bumpy transition, but there is every reason to think we can reduce the disruption to people’s lives and livelihoods,” citing society’s track record of managing risks from earlier technologies: “The best reason to believe that we can manage the risks is that we have done it before.” The new essay’s framing — one of the most turbulent times in human history — is a marked shift in tone from that 2023 assessment.

What we still do not know

Gates does not provide a tax rate, calculation method or enforcement mechanism for either tax, though he says a token tax should be targeted so it does not slow down purely beneficial uses of AI, such as medicine and education; no source found shows him responding to Etzioni’s criticism that it targets the wrong metric. He also raises, without answering, who would decide which jobs count as “Human Reserved,” what criteria would apply, how compliance would be enforced against companies substituting robots anyway, and how international trade would treat countries with different rules. No source found quantifies the extent of Gates’s current financial ties to Microsoft or other AI companies beyond his own statement that he has diversified but not eliminated them. Finally, no source found establishes whether any government has acted on Gates’s tax proposal since he first raised it years ago.