August 29, 2026
Bill Gates has done something more consequential than publish another forecast about artificial intelligence: he has set a table.
A table is an ordinary thing—wood, weight, empty chairs—until something critical is being decided around it. Then every chair matters, including the empty ones. Gates does not reserve this table for those who build or own the machines. He calls governments, companies, workers, educators, health professionals, communities, and religious leaders to sit together while meaningful choices remain. When much of the AI industry asks how fast and far we can go, he asks where we are going—and who may be left behind.
In “The Turbulent AI Era Is Here. The Choices We Make Now Are Critical,” Gates treats AI as a test of how we understand ourselves, form relationships, value work, and distribute power. AI could extend medicine, education, and opportunity, yet also weaken human relationships, displace workers, empower bad actors, and deepen inequality.
His manifesto deserves substantial credit. Gates speaks as an architect and beneficiary of the digital age, candidly acknowledging his ties to the industry. Yet he does not claim authority to settle these questions alone. He enlarges the circle of responsibility, admits what remains unanswered, and insists that democratic society—not technology companies alone—must govern the transition. This is civic leadership at its best.
Gates adds his name to the history of people who looked beyond the machinery of progress and seen what hangs in the balance: the fate of civilization, and whether our inventions will enlarge the human future or consume it. At the dawn of cybernetics, Norbert Wiener’s The Human Use of Human Beings (1950) warned that automation could free people from labor or deepen dehumanization. In postwar Europe, Martin Heidegger’s “The Question Concerning Technology” (1954) argued that technology changes how we encounter our world. Amid nuclear danger, the Russell–Einstein Manifesto (1955) urged humanity to rise above its divisions before its inventions destroyed it. Confronting nuclear, ecological, and genetic risks, Hans Jonas’s The Imperative of Responsibility (1979) extended responsibility to posterity and planet. Gates brings that responsibility into the AI era.
As someone developing AI in the service of human flourishing, I receive his essay not as a call to restrain innovation. But as a summons to make it worthy of humanity, and of the future we are now placing in its hands. Today’s choices will become the inheritance of generations that had no voice in making them. Gates has invited us to the table. We should answer with equal seriousness.
What Gates Gets Right
Only what, exactly, are we building? Why are we so captivated by what AI may become, and so hesitant to ask what it may make of us? Or, to turn to another question: how can we watch such extraordinary power gather in the hands of a few corporations, while the future it will shape belongs to all of humanity? Gates’s contribution is to insist that these questions can no longer be left at the margins. Six insights stand out.
1. Names the magnitude
AI could transform work, power, relationships, and human agency. Leaders must prepare before disruption becomes crisis.
2. Balances promise and peril
Gates avoids both utopianism and fatalism. Innovation and protection must advance together.
3. Centers equity
He asks who will benefit and who may be left behind. AI must be judged by its effects on the vulnerable, not productivity alone.
4. Defends meaningful work
Work can provide belonging, achievement, and worth—not simply income. Responses to displacement must address meaning as well as security.
5. Protects human presence
Healthcare, teaching, and moral judgment possess values efficiency cannot measure. Technical capability does not create moral permission.
6. Calls society to govern
Technology companies should not determine our future alone. Political, business, religious, and civic leaders must convene for the common good.
These contributions give public language to shared concerns and direct them toward action in service of worthy ideals. His warnings still require the same scrutiny as his hopes.
Where the Evidence Remains Unsettled
To support his concerns about employment, Gates cites an updated Stanford Digital Economy Lab study. The researchers found no evidence that AI is causing widespread job losses across the economy. But they did identify a troubling pattern among workers aged 22 to 25: employment in occupations highly exposed to AI was 19 percent lower than it would have been if those jobs had grown at the same rate as less-exposed jobs. The gap resulted mainly from employers hiring fewer young workers—not from firing people already employed.
To me, this is where a labor statistic becomes a human story. AI may not be eliminating work across the economy, but it may be weakening the first rung of the learning ladder—the junior roles through which young people acquire experience, develop judgment, and find their place in professional life. Still, the researchers describe these findings as early indicators, not causal estimates. Exposure is not elimination; neither is it harmless.
History offers no automatic reassurance. In The Technology Trap, Carl Benedikt Frey distinguishes labor-enabling technologies from labor-replacing ones. Industrialization eventually created enormous wealth, but many workers experienced only displacement, declining status, and political upheaval during their lifetimes. For them, the long run arrived too late.
Something similar unsettles the evidence on human well-being. A study of 1,131 adult AI-companion users associated companionship-oriented chatbot use with lower well-being, particularly when interactions became intensive and emotionally personal, and users lacked strong human support. But the direction remains unclear. AI companions may deepen isolation; people already experiencing isolation may turn to them; or each may reinforce the other. Gates is right to say that the evidence remains limited and mixed.
I do not read these studies as proof of inevitable catastrophe. But neither can I dismiss them as background noise. For now, we stand in an uneasy middle distance: between innovation that enlarges human possibility and innovation that quietly erodes the places where people learn to work, to relate, and to become themselves. Vigilance, then, is not fatalism. It is responsibility before certainty arrives.
Equity Is Necessary—but Insufficient
Gates rightly wants advanced intelligence to reach rural hospitals, poorer farmers, underserved students, and small organizations. But access is not flourishing. I use four complementary tests:
Ability — What can it do?
Can the AI perform the needed task reliably and responsibly in real-world use?
Access — Who can use it?
Who can obtain, afford, and benefit from it?
Agency — Who can shape it?
Do people who work with it have a meaningful voice in its purpose, design, and rules?
Authority — Who controls it?
Who owns it and has final power to govern, redirect, restrict, or stop it?
A worker may use AI yet have no say in how it monitors or reorganizes the job. A patient may receive an AI-supported diagnosis without knowing who remains answerable when it fails. And AI can appear widely available while control remains concentrated across its chips, cloud infrastructure, models, and platforms.
Gates sets the table for democratic deliberation. But a seat without decision rights is consultation, not governance. Authority over AI should belong to no company, government, or nation alone. It must operate at three levels: an international commission for shared standards and cross-border risks; national institutions for enforceable rights and duties; and workplaces and communities where people can shape, challenge, redirect, or stop systems operating in their lives.
A flourishing technology must expand technical ability and access alongside human agency and legitimate authority.
The Limits of “Human Reserved”
Gates’s most compelling proposal begins with memory. He recalls the caregivers who supported his father through Alzheimer’s disease—people whose attentiveness and presence could not be reduced to completed tasks. Gates calls his proposed domain Human Reserved: work and activities society would deliberately keep in human hands even when AI or robots could perform them. Like a nature reserve, it protects something not because development is impossible, but because the loss would be too great.
I find the metaphor imaginative and morally serious. Gates acknowledges that its boundaries would vary across societies, evolve over time, and ultimately require public deliberation. Just because a machine can do something, does that mean it should?
But every reserve has a boundary, and every boundary leaves something outside.
Imagine a physician present while AI determines diagnosis and treatment, or a teacher in the classroom while software sets content and assessment. Are they still practicing a vocation, or merely ratifying the machine’s work? At what point does presence become performance? At what point is the person no longer the author of the work, but merely the living signature beneath a decision already made?
John Danaher and Sven Nyholm describe this as an “achievement gap”: automation can diminish the meaningful effort through which people exercise abilities and experience accomplishment. A human name on the payroll is not evidence of human agency. “Human Reserved” protects valuable islands of human presence; it does not govern the ocean of AI-mediated life around them.
From Human Reserved to Human Authored
I would not discard Gates’s proposal. I would critique and extend it. In my TEDxPrinceton talk, “The Human Economy in the Age of AI,” I named a more hopeful possibility: “This is not the end of human relevance. This is the beginning of human renaissance.”
If machines increasingly assume what is replicable and transactional, humans can invest more deeply in what is relational, creative, moral, and transformative. But renaissance is not automatic. It requires more than preserving human presence inside automated institutions. It requires keeping our institutions—and our future—human-authored.
Human Reserved identifies activities machines should not assume. Human Authored establishes the governing principle wherever AI is present: machines may extend our capabilities, but humans must remain the source of meaning, judgment, responsibility, and direction.
I propose keeping humanity at the CORE:
C — Coauthorship
People and communities coauthor AI’s purposes, priorities, and boundaries.
O — Origin
Humans remain the source of purpose, judgment, creativity, and meaningful action.
R — Responsibility
Consequential decisions remain understandable, contestable, and answerable to people.
E — Equity
Authority, opportunity, and economic gains are shared broadly and fairly.
CORE does not preserve a secondary place for people inside machine-authored systems. It keeps machines as instruments within a human-authored society—consistent with human-centered AI research that seeks to augment and empower people rather than diminish or replace them.
A Future Humanity Can Call Its Own
Gates is right: this future is not predetermined. He has called us to the table. Now leaders must build institutions capable of governing AI for the common good.
This is where I would go further than Gates: preserving human spaces is not enough if we surrender authorship of the world around them. Humanity should not become a protected species occupying a few carefully preserved enclaves within a machine-governed world. I do not want humanity preserved like an artifact behind glass, visible, honored, and powerless. I want us alive in the authorship of the world: naming its purposes, exercising its judgment, bearing its responsibility, and sharing its prosperity.
Intelligence may be artificial. Authorship must remain human.