In Rome, a gathering of Nobel laureates has issued a fresh call for the world to confront two existential threats at once: nuclear weapons and the rapid advance of artificial intelligence. Backed, it is suggested, by the moral authority of the Pope himself, the declaration urges global efforts to "disarm" AI, to constrain its development, limit its military applications, and prevent it from becoming another vector for human destruction.
It is a noble sentiment. It will almost certainly fail. The gathering reflects an understandable anxiety. Artificial intelligence is no longer a distant prospect confined to research labs. It is being integrated into military planning, autonomous systems, cyber operations, and decision-making loops at a pace that outstrips international norms and treaties. Combined with the persistent shadow of nuclear arsenals, the image of AI-augmented doomsday scenarios feels uncomfortably plausible. Who could object to wise voices calling for restraint?
Yet the deeper reality is harsher. AI is not a weapon that can be locked away in an arsenal or dismantled like a missile silo. It is not a physical object that nations can agree to ban and then verify compliance through inspections. Artificial intelligence is a general-purpose technology, a new form of intelligence that emerges from code, data, and computation. It spreads through software updates, research papers, open-source repositories, and commercial competition. Once the underlying capabilities exist, they cannot be uninvented.
History offers little comfort to those hoping for successful disarmament. Nuclear weapons, for all the treaties and non-proliferation efforts, remain the ultimate guarantor of great power sovereignty. Nations that possess them guard them jealously. Those that do not, pursue them in secret or under civilian cover. The same incentives that make nuclear disarmament nearly impossible apply with even greater force to AI. The technology is dual-use by nature. The economic and military advantages of leading in AI are so overwhelming that no major power, the United States, China, Russia, or others, will voluntarily cripple its own capabilities while rivals advance.
Even if the Pope lends his considerable moral weight to the cause, the limits of moral suasion in great power competition are well known. Nations do not disarm because they are scolded or inspired by ethical declarations. They disarm when they calculate that doing so enhances their security or when they have no choice. Neither condition currently exists for AI.
The declaration also underestimates how deeply AI is already embedded in modern life. From logistics and medicine to scientific research and financial systems, the technology delivers tangible benefits that societies are unwilling to forgo. Attempts to "disarm" AI risk creating a world where only authoritarian regimes or bad actors ignore the rules, while democratic societies hobble themselves. The result would not be peace, but a dangerous asymmetry that could make conflict more likely.
This is the uncomfortable truth of technological progress: once a powerful capability is discovered, it tends to spread. The genie does not return to the bottle because laureates and religious leaders ask politely. AI will not pack its bags and leave simply because we fear its potential for harm. It will be shaped, regulated, and contested, but not eliminated.
The wiser path lies not in grand declarations of disarmament, but in hard-headed efforts to build resilience, maintain strategic advantage where it matters most, and develop robust norms around the most dangerous applications. Moral voices have their place in reminding us of our humanity. But they cannot repeal the realities of power, competition, and the irreversible momentum of technological change.
The world, as usual, will continue marching forward on its own terms. AI is here to stay. The only question is whether we will master it, or allow it to master us.