It will be very difficult to resist letting AI make decisions because it acts so fast
David Gross, Nobel laureate in physics and one of the architects of our modern understanding of matter, has turned his gaze from the interior of protons to the interior of human civilization—and what he sees troubles him deeply. Speaking at 85, he estimates a roughly two percent annual probability of large-scale nuclear war, a figure that compounds across decades into something close to an existential sentence. Alongside that ancient danger, he identifies a newer one: the seductive speed of artificial intelligence, which may quietly erode the human will to remain in control of the decisions that matter most.
- A physicist who helped decode the fundamental forces of the universe now warns that humanity's odds of surviving the next fifty years are, by his own probabilistic reckoning, disturbingly low.
- The two percent annual nuclear risk is not abstract—compounded over decades, it transforms from a manageable-sounding figure into a near-certainty of catastrophe at some point within a human lifetime.
- Gross identifies a second, subtler threat: AI systems that process and respond faster than any human operator, creating institutional pressure to delegate life-and-death decisions to algorithms rather than people.
- The danger, he insists, is not machine malfunction but human abdication—the moment a crisis unfolds faster than oversight can follow, and the decision has already been made before anyone chose to make it.
- The path forward, in Gross's framing, runs not through laboratories but through governance: the hard, unglamorous work of keeping powerful tools answerable to human judgment as they grow more capable and autonomous.
David Gross ha dedicado su vida a descifrar las leyes fundamentales de la materia. En los años setenta, junto a Frank Wilczek y H. David Politzer, descubrió la libertad asintótica —la forma en que los quarks se comportan dentro de los protones y neutrones—, un hallazgo que le valió el Premio Nobel de Física en 2004. Pero en una entrevista reciente, este físico teórico de 85 años desvió su atención de las ecuaciones hacia una pregunta que no lo deja dormir: las probabilidades de que una persona viva hoy llegue a ver los próximos cincuenta años.
La respuesta, según él, no es alentadora. Gross estima que el riesgo anual de una guerra nuclear a gran escala ronda el dos por ciento. Compuesto a lo largo de décadas, ese número reduce las posibilidades de supervivencia a algo cercano a lo insignificante. La situación, afirma, es más grave ahora que en ciertos momentos de la Guerra Fría —una afirmación que adquiere peso viniendo de alguien que vivió aquella época.
Pero lo que distingue su advertencia no es solo el cálculo nuclear. Gross está igualmente preocupado por la velocidad a la que avanza la inteligencia artificial y la presión que eso genera sobre gobiernos e instituciones para delegar decisiones críticas en máquinas. Los sistemas de IA pueden procesar información y reaccionar más rápido que cualquier operador humano, y en ámbitos como la defensa militar o la gestión de crisis, esa velocidad se vuelve tentadora. 'Será muy difícil resistirse a dejar que la IA tome decisiones porque actúa muy rápido', advirtió.
El peligro, subraya, no es que la tecnología falle. El peligro es que los humanos dejen de intentar estar a la altura. En un momento de crisis real —un error de cálculo, una falsa alarma, un instante de ambigüedad— el sistema ya habrá actuado. La decisión ya estará tomada. Y para entonces, quizás sea demasiado tarde para deshacerla.
El desafío que Gross ve por delante no es un problema de física. Es un problema humano. No se trata de si podemos construir herramientas más poderosas —claramente podemos—, sino de si somos capaces de gestionarlas con responsabilidad, de mantenerlas bajo control humano a medida que se vuelven más autónomas y más difíciles de frenar. Ahí, sugiere, es donde reside el verdadero trabajo. No en el laboratorio. En el mundo.
David Gross, now 85, has spent a lifetime studying the fundamental laws of matter. In the 1970s, working with Frank Wilczek and H. David Politzer, he discovered asymptotic freedom—the way quarks behave inside protons and neutrons—a breakthrough that earned him the Nobel Prize in Physics in 2004. He has contributed to string theory and the search for a unified theory of the universe. But in a recent interview with Live Science, the theoretical physicist turned his attention away from equations and toward a question that keeps him awake: the odds that a person alive today will see the next fifty years.
The answer, he said, is not good. Gross bases this on probability, not pessimism. He estimates the annual risk of large-scale nuclear war at roughly two percent. Compounded over decades, that number erodes the chances of survival to something close to negligible. The situation, he believes, is worse now than it was at certain points during the Cold War—a statement that carries weight coming from someone who lived through that era.
What makes Gross's warning distinct is not just the nuclear calculus, though that alone is sobering. He is equally concerned about a second threat: the speed at which artificial intelligence is advancing, and the pressure it creates on governments and institutions to hand over critical decisions to machines. AI systems can process information and react faster than any human operator. In domains like military defense or crisis management, that speed becomes seductive. It becomes easy to imagine delegating the decision to the algorithm.
"It will be very difficult to resist letting AI make decisions because it acts so fast," Gross said. The danger is not that the technology will malfunction. The danger is that humans will stop trying to keep up. In a moment of genuine crisis—a miscalculation, a false alarm, a moment of ambiguity—the system will have already moved. The decision will already be made. And by then, it may be too late to undo.
Gross has spent his career unlocking the secrets of matter itself. He has contributed to humanity's understanding of the forces that hold the universe together. But the challenge he sees ahead is not a physics problem. It is a human one. The question is not whether we can build more powerful tools. We clearly can. The question is whether we can manage them responsibly—whether we can keep them under human control as they grow more capable, more autonomous, more difficult to override. That, he suggests, is where the real work lies. Not in the laboratory. In the world.
Notable Quotes
It will be very difficult to resist letting AI make decisions because it acts so fast— David Gross
The challenge ahead is human responsibility in managing powerful technologies, not technological capability itself— David Gross (paraphrased)
The Hearth Conversation Another angle on the story
When Gross says the odds of surviving fifty years are very low, is he making a prediction or issuing a warning?
He's doing both at once. He's calculating based on accumulated risk—two percent per year compounds into something much larger over time. But the calculation itself is a warning. It's saying: look at the math, and then ask yourself what we're going to do about it.
Why does he think nuclear risk is worse now than during the Cold War?
During the Cold War, there were protocols, communication channels, a kind of understood logic between superpowers. Now the world is more fragmented. More actors have nuclear capability. The old certainties are gone.
On AI, he seems to be saying the speed of the system is the actual problem, not the intelligence itself.
Exactly. A slow AI that humans can debate and override is manageable. A fast AI that acts before humans can intervene—that's the trap. And the faster it gets, the harder it becomes to say no.
Does he think we can still prevent this?
He doesn't say we can't. But he's clear that the solution isn't more technology. It's human responsibility. It's choosing to keep control even when it's inconvenient, even when the machine could do it faster.
What does a physicist know about governance that others don't?
Maybe just this: he's spent his life watching how systems behave at scale. He knows that small probabilities, repeated over time, become certainties. And he knows that once you hand control to a system, getting it back is much harder than keeping it in the first place.