We are in the Stone Age of AI, and it is already reshaping society.
UN creates unprecedented global scientific body to evaluate AI's transformative impact and guide international governance before development outpaces regulation. Spanish expert Román Orús compares AI development to the Manhattan Project, emphasizing need for control mechanisms despite current systems remaining in early, inefficient stages.
- UN creates first global scientific panel on AI with 40 experts from 140+ countries
- Spanish physicist Román Orús is the sole Spanish representative on the panel
- First in-person meeting held April 22-24, 2026 in Madrid
- Panel co-chaired by Turing Prize winner Yoshua Bengio and Nobel laureate Maria Ressa
- Over 2,600 applications received for 40 positions
The UN inaugurates its first independent scientific panel on AI in Madrid with 40 experts, including Spanish physicist Román Orús, who warns the technology requires control frameworks comparable to nuclear energy management.
The United Nations has taken an unprecedented step. On April 22, it will convene the first meeting of its International Independent Scientific Panel on Artificial Intelligence in Madrid—a body born from the recognition that AI has stopped being a distant technological promise and become a force actively reshaping economies, political systems, and the daily lives of millions. The panel, approved by the UN General Assembly in August 2025, exists because of a growing international anxiety: how do you govern a technology that evolves faster than regulation, with transformative power comparable to the great industrial revolutions, but moving at a pace no single country or company can fully anticipate?
Forty experts from more than 140 countries will form this panel, selected from over 2,600 applications through an open international call. They represent academia, private industry, civil society, and international organizations. Nineteen are women, twenty-one are men. They are co-chaired by Yoshua Bengio, a Turing Prize winner, and Maria Ressa, a Filipino journalist and Nobel Peace Prize laureate. Each member acts independently, required to disclose conflicts of interest—a safeguard the UN has built in to protect the panel's scientific credibility at a moment when trust in AI governance is fragile. Among the forty sits a single Spanish representative: Román Orús, a physicist specializing in quantum computing and a recognized figure at the intersection of artificial intelligence and emerging technologies.
Orús directs the scientific work at Multiverse Computing and holds a research professorship at the Donostia International Physics Center in San Sebastián. His appointment reflects the panel's hybrid ambition—to blend academic research with industrial application. But beyond his institutional role, Orús offers a vision of AI's current moment that is notably sharp and sometimes unsparing. He compares the technology's development to the Manhattan Project, one of the twentieth century's most consequential episodes. The comparison is deliberate. Just as with nuclear energy, the question is not merely technological potential but the necessity of establishing limits and control frameworks to prevent unintended consequences. "The AI is almost the new 'Manhattan Project'. And when a technology of this caliber and with this capacity emerges, you cannot simply leave it loose," he has said.
Yet Orús introduces a qualification that often disappears from more optimistic discourse. Despite the advances, artificial intelligence remains in an extremely early phase. "We are in the Stone Age of AI," he warns. Current systems, however sophisticated they appear, are still limited, inefficient, and staggeringly expensive in energy terms. To illustrate the paradox, he offers a comparison: the models we have are like the chariots that existed in ancient Egypt thousands of years before Christ, compared to the starship Enterprise. Much work remains in both software and hardware development. Biological intelligence, he notes, remains infinitely more efficient. A child learns to speak without being placed beside a nuclear power plant. An AI model, by contrast, might require more than a billion euros in electricity costs and still produce hallucinations.
The panel's mandate is to produce annual reports with evidence-based scientific assessments of AI's opportunities, risks, and impacts. It aspires to become a bridge between scientific knowledge and political decision-making, guiding possible international AI governance. UN Secretary-General António Guterres framed it in foundational terms during his address to the panel members: "a global scientific body as has not existed before," called to help "shape the development of artificial intelligence for the benefit of humanity, while we still have time." He emphasized the velocity of change: "Never in the future will we advance as slowly as we are advancing now."
Madrid was chosen as the site for this first in-person meeting—held April 22 to 24—partly because Spain, alongside Costa Rica, co-facilitated the intergovernmental process that created the panel. Spain has intensified its commitment to international AI governance in recent months, including plans to host the UN's AI Governance for Humanity Lab in Valencia, funded by the Spanish government. The panel will address employment impacts, applications in healthcare, ethical risks tied to generative models, and mechanisms to distinguish human-created from machine-created content. Underlying all of this is a central principle the UN has repeatedly emphasized: artificial intelligence must expand human capabilities, not replace them. The panel's work, beginning in Madrid this week, represents the world's first serious institutional attempt to ensure that distinction holds.
Citas Notables
The AI is almost the new Manhattan Project. When a technology of this caliber emerges, you cannot simply leave it loose.— Román Orús, Spanish physicist and panel member
A global scientific body as has not existed before, called to shape the development of artificial intelligence for the benefit of humanity, while we still have time.— UN Secretary-General António Guterres
La Conversación del Hearth Otra perspectiva de la historia
Why does the UN feel it needs this panel now, specifically? Why not five years ago, or why wait five more years?
Because the speed changed. ChatGPT arrived and suddenly AI wasn't theoretical anymore—it was in people's hands, reshaping labor markets and politics in real time. Regulation was nowhere near ready. The UN realized that if they waited, the technology would have already written the rules.
Román Orús compares it to the Manhattan Project. That's a heavy comparison. What does he mean by it?
He means that some technologies are too powerful to let develop without guardrails. With nuclear energy, we learned that lesson the hard way. With AI, we have a chance to learn it before the consequences are irreversible. You can't just let it loose and hope for the best.
But he also says we're still in the Stone Age of AI. If we're so early, why the urgency?
That's the paradox. We're early, yes—the systems are inefficient, they hallucinate, they cost billions to train. But they're already reshaping society. Imagine what happens when they actually work well. The time to set boundaries is before the technology becomes indispensable.
What does the panel actually do? Does it have enforcement power?
No. It produces annual reports with scientific assessments. It's meant to inform policy, not make it. The real power is in credibility—if forty of the world's leading minds on AI agree on something, governments have to listen.
Why is it significant that there's only one Spanish expert on a forty-person panel?
It shows how concentrated expertise still is globally. Spain has real talent in this space, but the field is dominated by a handful of countries. That imbalance itself is something the panel will have to reckon with.