Three years ago, a few months after ChatGPT was released, we wrote in this newspaper suggesting that the G20 charter an international panel on technological change. The purpose would have been to help shepherd humanity through what was already becoming a turbulent period, as advances in Artificial Intelligence (AI) and biology began to leave laboratories and become powerful forces in human affairs.
Our reasoning was straightforward. Before anything effective could be done, the major economic and scientific powers needed a shared understanding of what was happening. Political leaders required relatively direct and frequent access to the best available analysis of emerging technologies, their risks, and their opportunities. The G20 seemed the appropriate forum: Large enough to include the countries that matter, but not so large as to be unwieldy, and not so politically narrow as to exclude essential actors such as China. A body that was too expansive would be ineffective; one defined by political affinity would defeat the purpose of seeking species-wide coordination.
Since then, activity has accelerated. A summit process has emerged and continues later this week in New Delhi. Yet the type of standing, structured mechanism we proposed does not exist. We fear that the political leaderships of major countries remain far from having optimal arrangements in place to understand and manage developments that are advancing at extraordinary speed.
The pace of change has been unmistakable. AI has been highly visible, not least because millions of people are already using AI systems and because enormous investment has flowed into the sector. But chatbots are only the most familiar face of the transformation. Autonomous weapons have been deployed in active conflict. Distributed AI systems are proliferating beyond major research laboratories, and dedicated computational infrastructure continues to expand. Taken together, these developments begin to resemble the early stages of a new technological “biosphere”, within which processes increasingly unfold with limited human supervision or comprehension.
On the biological front, companies offering genetic selection services to prospective parents are now operating. If such practices become widespread, they could generate cohort effects in which average characteristics of birth cohorts shift in systematic ways. Meanwhile, AI is accelerating biological research itself, compressing timelines for discovery and intervention. The convergence of these domains has only begun.
We do not raise these examples to pass moral judgment. As scientists, we recognize that evolution did not stop with the emergence of Homo sapiensand it will not end with humans as presently constituted. What concerns us is not change per sebut the timescale on which change is occurring. The speed of transition matters.
In our earlier essay, we suggested that the long-standing “human equilibrium” might be ending. Today, that claim appears less speculative. Many of our political, economic, and social arrangements have become metastable. A familiar physical analogy is water heated beyond its boiling point yet remaining liquid — a temporary state that can give way abruptly to steam.
Consider several domains. Large numbers of economic roles rely on cognitive skills within bounded domains of knowledge; in principle, many are automatable. Emerging AI systems embedded in wearable devices could mediate social interaction in ways that fundamentally alter how humans relate to one another, shifting from intuitive engagement to continuous prediction and monitoring of internal states.
Parents struggle to advise children on education or careers because the time required to reach maturity now exceeds the projected timeline of major technological upheavals. And reproductive technologies introduce the possibility that parents select embryos on the basis of polygenic traits, reshaping human characteristics over generations.
To many people, daily life appears unchanged. Yet structurally, everything has changed. The concern is that metastable systems do not necessarily transition smoothly. They can undergo abrupt phase shifts — politically, economically, socially — that produce instability before a new equilibrium forms. Technological acceleration may generate explosive adjustments rather than gradual adaptation.
Perhaps this concern is misplaced. Perhaps technological enthusiasts and accelerationists are correct that rapid development will produce severe net benefits without disruption. We would simply prefer that such optimism be tested against systematic analysis rather than assumed. Political decision-making should be informed by the best available interdisciplinary understanding of risks, trajectories, and systemic effects.
Here we encounter a structural problem. In a competitive commercial environment, firms have little incentive to slow development. Scientists and engineers operating within that ecosystem face similar pressures.
At the international level, if political leaders lack a deep understanding of technological dynamics, it becomes easier to default to nationalist or economic framings. Under such conditions, coordination is difficult, and mutual suspicion grows.
Given the current geopolitical climate, it may be unrealistic to renew our earlier call for a formal international panel on technological change. Instead, we propose something more modest and potentially more feasible.
Major governments should appoint, or pivot, senior science and technology advisors whose mandate goes beyond traditional research administration, and into ongoing analysis and forecasting of transformative technologies. These advisors should be comparable to that of national security advisors. Political leaders should commit to regular, structured briefings, monthly, for example, focused specifically on technological trajectories, systemic risks, and strategic implications.
The goal is not to centralize control or to impose global uniformity. It is to raise the salience of technological acceleration at the highest levels of decision-making. If such advisory structures were established across major countries, a community of senior analysts would naturally emerge.
Over time, they might conclude that certain categories of risk, especially those that are genuinely catastrophic or existential, require at least partial coordination.
We emphasize that this is not an argument for world government. As citizens of open societies, we value our own political and economic institutions. Democracies and autocracies will disagree on many issues, and such disagreement is not going away. But coordination failures can occur even among rivals when shared risks are misunderstood or ignored.
The Cold War never escalated into nuclear war, not because ideological conflict disappeared, but because a basic interest in survival was recognized across political systems. Today’s technological moment is more complex. AI and biotechnology are diffuse, commercially embedded, and capable of delivering immense civilian benefits. They are not confined to state arsenals. That very diffuseness makes governance more challenging. Yet difficulty does not absolve responsibility.
We are entering a period in biological, technological and cultural evolutionary processes that may operate on timescales far shorter than those to which human institutions are adapted. The question is not whether change will occur. It will. The question is whether political leadership will be systematically informed about the pace, direction, and implications of that change.
A high-level, recurring technological briefing embedded within national decision-making, would not solve every problem. This is a modest institutional reform, but it would improve the quality of strategic awareness. And improved awareness is a prerequisite for any meaningful coordination. We hope that, in their deliberations, participants at the current summit will consider steps in this direction.
Martin Rees is former president of the Royal Society and a member of the House of Lords; Shivaji Sondhi is Wykeham professor of physics at Oxford and professor emeritus at Princeton; and K VijayRaghavan is former principal scientific advisor to the government of India. All three authors are fellows of the Royal Society. The views expressed are personal
