In the quiet hours when the mind turns toward horizons beyond the self, humanity now confronts not one shadow but many — artificial intelligence among them — each capable, in theory, of foreclosing the human story entirely. What distinguishes this moment from prior eras of civilizational anxiety is not the presence of a single, legible threat but the simultaneous emergence of many, arriving faster than our institutions or our psychologies were built to absorb. The deeper challenge is not merely technical but profoundly human: how does a species wired for immediate danger learn to think clearly
Weighing A.I.'s Existential Risk Among Humanity's Cosmic Fears
The human brain evolved to handle immediate threats, not statistical ones.
Mark
So the piece is saying that AI is just one threat among many. But doesn't that actually make it harder to focus on any single one?
Mimi
Exactly. If you're a policymaker or a researcher, you're being asked to attend to AI risk, pandemic risk, nuclear risk, climate risk—all at once, all potentially civilization-ending. The cognitive load is real.
Luke
But we should be careful here. The piece doesn't actually quantify how serious any of these threats are relative to each other. It says they're all "real" and "studied by serious people," but it doesn't tell us whether AI risk is 10 percent of the total existential risk or 50 percent.
Mimi
That's fair. The piece is more about the psychological and institutional challenge than about comparative risk assessment.
Mark
Is there a sense in the reporting of which threat is getting the most attention right now?
Mimi
AI seems to be getting a lot of oxygen in the room, but the piece doesn't make a strong claim about whether that's proportional to the actual risk.
Luke
Right. And that's important, because if we're allocating resources, we need to know whether we're over-investing in AI safety relative to, say, biosecurity. The piece identifies the problem but doesn't resolve it.
Mark
So what's the actual forward-looking question here?
Mimi
How do we build institutions and ways of thinking that can handle multiple existential threats without either paralysis or denial?
Luke
Which is a good question, but the piece doesn't really show us what that looks like in practice. It's more diagnosis than prescription.
O Pulso
- AI has joined a crowded field of existential threats — engineered pandemics, nuclear war, climate collapse, synthetic biology — that serious researchers now treat not as science fiction but as policy emergencies.
- The human brain, evolved for visible and immediate danger, strains under the cognitive weight of threats that are abstract, probabilistic, and measured in centuries rather than seconds.
- Comparing risks creates its own vertigo: AI is being actively built and deployed, while other catastrophic possibilities operate on different timelines and through entirely different mechanisms, making rational prioritization feel nearly impossible.
- The psychological toll is underexamined — paralysis and denial compete with the urgent need for clear-headed decision-making about research priorities, resource allocation, and governance.
- What is taking shape is a recognition that entirely new cognitive and institutional frameworks must be built — ones capable of holding multiple, overlapping, sometimes contradictory existential threats in view at once without collapsing into despair or false optimism.
In the quiet hours when the mind turns toward horizons beyond the self, humanity now confronts not one shadow but many — artificial intelligence among them — each capable, in theory, of foreclosing the human story entirely. What distinguishes this moment from prior eras of civilizational anxiety is not the presence of a single, legible threat but the simultaneous emergence of many, arriving faster than our institutions or our psychologies were built to absorb. The deeper challenge is not merely technical but profoundly human: how does a species wired for immediate danger learn to think clearly about abstract, statistical, century-scale catastrophe?
Somewhere in the middle of the night, the mind drifts toward the end of things — not personal endings, but the civilizational kind. Artificial intelligence has become the latest focal point of that anxiety, a technology powerful enough and opaque enough that serious researchers now spend their careers mapping the ways it could go irreversibly wrong.
But AI does not occupy this territory alone. It shares it with engineered pandemics, nuclear arsenals, climate tipping points, and synthetic biology experiments that might escape containment. The study of these possibilities has become its own academic field, complete with funding, conferences, and a growing community of people who treat the end of civilization as a subject worthy of rigorous, sustained attention.
What makes this genuinely hard — not just intellectually but psychologically — is the mismatch between the scale of these threats and the brain that must process them. Human cognition evolved for immediate, tangible danger. Existential risk asks something different: to think in probabilities and centuries, to weigh scenarios that may never materialize against the overwhelming realness of the present moment.
Comparison compounds the difficulty. How does one weigh AI risk against the risk of a bioengineered pathogen? Both are taken seriously by serious people. Both could, in principle, end civilization. Yet they operate on different timelines, through different mechanisms, and with different degrees of human agency in their prevention. The act of comparison itself produces a kind of vertigo.
This is not the first generation to live with knowledge of catastrophic possibility — the Cold War and the slow unfolding of climate change are recent precedents. But what distinguishes the present moment is the velocity of change and the sheer number of threats that have migrated simultaneously from science fiction into policy debate. The challenge is no longer choosing between existential risks but building institutions capable of attending to all of them at once.
No clear answer has emerged from this landscape — only a recognition. The future will demand new cognitive and institutional tools: frameworks that take AI seriously without dismissing other dangers, ways of thinking that resist both techno-optimism and apocalyptic despair. The work has begun. It is far from finished.
Somewhere in the middle of the night, you might find yourself thinking about the end of things. Not the end of your day, or your year, but the end—the scenario in which human civilization as we know it simply ceases. Artificial intelligence has become the latest candidate for this role in our collective anxiety, a technology so powerful and so poorly understood that serious researchers and policymakers now spend their days gaming out scenarios where it goes wrong in ways we cannot predict or control.
But here is the thing that makes the modern moment genuinely strange: artificial intelligence is not alone in this category of threat. It shares company with a roster of other possibilities that could, in theory, end us all. Engineered pandemics. Nuclear war. Asteroid impacts. Climate cascades that trigger systems collapse. Synthetic biology experiments that escape containment. The list is long enough that it has become its own field of study, complete with academic chairs, funding streams, and conferences where people gather to discuss, with varying degrees of alarm, the many ways the future might not arrive.
What makes this exercise genuinely difficult—not just intellectually but psychologically—is the sheer cognitive burden of holding all of these possibilities in mind at once. The human brain evolved to handle immediate threats: the predator in the bush, the rival at the boundary of the tribe, the storm on the horizon. We are equipped with fear responses calibrated to things we can see and touch and run from. Existential risk operates at a different scale entirely. It is abstract. It is statistical. It is the kind of thing that requires you to think in probabilities and centuries, to weigh scenarios that may never happen against the certainty of the present moment, which always feels more real than the future, no matter how carefully you calculate.
There is also the problem of comparison. If you accept that artificial intelligence poses some meaningful risk to human survival, how do you weigh that against the risk posed by, say, a bioengineered pathogen? Both are real. Both are studied by serious people. Both could, in principle, end civilization. But they operate on different timelines, with different mechanisms of failure, and with different degrees of human agency in their prevention. One is a technology we are actively building and deploying. The other is a capability that exists in the realm of theoretical possibility, constrained by biology and access to resources. The comparison itself becomes a kind of vertigo.
This is not a new problem in human thought. Every generation has faced the question of how to live with knowledge of catastrophic possibility. The Cold War gave us nuclear anxiety. Climate science gave us the slow-motion catastrophe of planetary warming. But what is different now is the velocity of change and the sheer number of simultaneous threats that have moved from the realm of science fiction into the realm of serious policy discussion. We are not choosing between one existential risk and another. We are trying to build institutions and frameworks that can attend to multiple, overlapping, sometimes contradictory threats at the same time.
The psychological challenge is real and underexamined. How do you maintain focus and motivation when the stakes are framed as existential? How do you avoid either paralysis or denial? How do you make decisions about resource allocation, research priorities, and policy when the underlying uncertainties are so profound? These are not merely technical questions. They are questions about how humans can think clearly when the subject matter is designed to overwhelm the capacity for clear thought.
What emerges from this landscape is not a clear answer but a recognition: the future will require us to develop new cognitive and institutional tools for processing multiple, large-scale threats simultaneously. We will need frameworks that allow us to take artificial intelligence seriously without dismissing other risks, and vice versa. We will need ways of thinking that do not collapse into either techno-optimism or apocalyptic despair. The work has begun, but it is far from finished. The mind-bending exercise continues.
Quer a matéria completa?
Leia o original em The New York Times ↗
Technology