The race for artificial intelligence has collided with geopolitical rivalry and the climate crisis. The real question is not how fast AI can and should advance, but whether humanity can govern its direction.

There is something profoundly curious happening in the artificial intelligence revolution. For years, the message from the world's technology frontier was simple: move faster. Bigger models. More computing power. More chips. More data. More investment. Whoever reached the frontier first would shape the future.

Now, suddenly, some of the very people who have driven that race are asking humanity to slow down. In September, Anthropic CEO Dario Amodei called for a deliberate slowing of frontier AI development, proposing stronger independent evaluation and international cooperation. OpenAI CEO Sam Altman and others subsequently expressed support for pacing the frontier.

Should we take these warnings seriously? Certainly. Should we also ask what happens when the people who build a technology become influential voices in deciding how fast that technology should scale and be regulated? Certainly. But there is a larger question that is being missed.

Why has the speed of AI become such an urgent question precisely when humanity is already engaged in devising ways to possibly deploy AI for confronting two priority crises: geopolitical fragmentation and climate change? The answer may determine whether AI becomes humanity's greatest multiplier, or another mechanism through which power becomes concentrated in fewer hands.

The Race Behind the Race

AI is commonly presented as a revolution in software. It is not. Behind every software system are semiconductor factories, data centres, electricity networks, cooling systems, water and critical rare-earth minerals. The weightless "cloud" rests upon a very physical geography of mines, factories, power stations and infrastructure.

This matters enormously for geopolitics. China's dominant position in rare-earth processing illustrates the vulnerability of the "inclusive" AI being promoted by the UN. The AI race is therefore not simply about who writes the most powerful algorithm. It is about who controls the minerals needed for chips, energy resources, talent and markets. The digital revolution is becoming a contest over physical resources as well as intellectual resources, much like what the world saw with the nuclear revolution before and during WWII and later.

This transforms the geopolitical equation. AI may be digital at the interface. Its foundations are physical. The cloud has a geology!

The Strange Politics of Slowing Down

This brings us back to the calls for restraint, for which there appear to be legitimate reasons. AI capabilities may be advancing faster than safety mechanisms, laws and institutions can adapt. Rules requiring expensive testing, auditing, licensing and computing infrastructure can be easier for technology giants to absorb than for small companies, universities or open-source developers. The possibility of regulation may become a new technological moat.

This does not prove that calls for safety are disguised attempts to protect market dominance. It does mean that AI governance cannot be designed exclusively by those with the greatest commercial or geopolitical stake in AI. This is particularly important for India and China, as well as the wider Global South.

Global South Cannot Remain an Audience

India possesses enormous pools of scientific and engineering talent, a vast digital population and bubbling AI ambitions. China possesses mines of rare-earth materials, extraordinary manufacturing capacity, critical-mineral processing capabilities and a powerful technological ecosystem that can dominate the world. Are "slow please, speed-breaker ahead" signs by the Global North emerging from these obvious realities?

That gives rise to the suggestion of strategic cooperation between India and China, "going beyond borders". There are obvious areas in which India and China could indeed cooperate. Yet such dreamy cooperation should not be confused with strategic alignment. The two countries continue to have substantial differences involving trust, security, technology and geopolitical influence.

The more realistic prospect is selective cooperation alongside strategic competition, a form of technological "co-opetition" and not "co-mpetition"! That is a "Third Way" for the Global South. India, for example, need not choose permanently between Washington and Beijing. It can work with both and depend exclusively on neither.

AI may enter warfare, but it also has the potential to make diplomacy more effective.

The more serious scenario is not about the speed of development but the scary geopolitical implications when AI enters warfare. AI can accelerate intelligence analysis, surveillance, targeting and military decision-making. If machines compress the time between identifying a threat and responding to it, they can also compress the time available for human judgement.

But there is another side to the story. AI can help diplomats analyse vast quantities of information, model possible settlement scenarios, monitor ceasefires and identify common ground among populations divided by conflict.

The UN-supported Libyan experience recently published is instructive. AI-assisted digital dialogue enabled mediators to process large numbers of citizen responses across political and regional divisions, helping reveal areas of consensus that conventional elite-led negotiations had struggled to capture. The lesson is important: AI did not make peace. People did. AI helped people hear one another. That should be the model for AI in diplomacy. 

Machines can become extraordinary diplomatic co-pilots.

And What About Climate Warfare?

Climate change presents perhaps the greatest test of what AI should be reoriented towards. Humanity needs faster mitigation, reversing the rising emissions curves, better climate forecasting, smarter electricity grids, resilient agriculture, more efficient industries, better disaster preparedness and accelerated scientific discovery. AI can contribute to all of them.

Specialised AI models can improve weather prediction, detect methane emissions from soil and refineries, balance electricity grids and demand, and accelerate the discovery of new materials. The timing could hardly be more consequential as the world seeks to accelerate Net Zero.

The World Meteorological Organization now reports that El Niño is firmly established and expected to strengthen, with an exceptionally high likelihood of persisting through February 2027. Such an event can alter rainfall and temperature patterns worldwide and risks devastation from floods, drought and extreme heat.

For a farmer facing drought, a hyper-local AI forecast delivered to a mobile phone, along with a list of actions needed, may matter more than the world's most sophisticated chatbot. For a city facing extreme heat, an intelligent electricity grid may matter more than another spectacular demonstration of generative AI. For a community facing floods, an early warning delivered hours earlier can save lives, rather than another fancy demonstration of male and female robots.

This suggests a different measure of AI progress. Not simply "slow down", but "upgrade"!

The AI Climate Paradox

There is, however, a warning embedded in this opportunity. AI itself requires enormous infrastructure. Data centres consume electricity and water. Their hardware requires minerals. If AI expands through increasingly resource-intensive systems powered by carbon-intensive energy, the technology intended to help solve climate change could simultaneously increase environmental pressures.

The AI revolution therefore needs its own environmental accounting. Major AI infrastructure should increasingly be evaluated not only by accuracy, speed and computational power, but also by energy consumption, water use, carbon footprint and material dependence.

This is where the concept of selective acceleration and slowing down becomes a reality. Green AI must move from the margins into mainstream technology policy with speed.

From Artificial Intelligence to Humanware

There is, however, an even deeper gap in today's debate that overrides the warnings about slowing down. We speak constantly about hardware and software. We speak much less about humanware!

Yet the future of AI will ultimately depend upon human capacity to use, question, govern and redirect technology. A farmer needs access to useful intelligence, not technological spectacle. A student needs capability, not dependency. A worker needs pathways to acquire new skills. A citizen needs transparency and agency, not invisible algorithms making consequential decisions.

A developing country needs technological sovereignty, not another cycle in which raw materials are extracted locally, advanced technology is developed elsewhere and dependence returns in digital form.

This is where universities and R&D institutes can become crucial catalysts. They should not merely teach students how to use AI. They can become living laboratories for AI, climate action and sustainable development, bringing together engineering, science, management, ethics, public policy and community knowledge.

Young people should not be trained merely to become better users of machines. They should become better stewards of the systems in which machines operate. That is Humanware!

A Prudent Forecast

Some corporations will pursue scale. Some regulators will pursue control. And some powerful enterprises will dictate speed control. Billions of ordinary people, as happened with mobile phones and apps, will increasingly discover that AI is not an external technology. It is becoming part of the infrastructure of everyday life.

The central danger is, therefore, not simply that AI will be monstrous if the accelerator is pressed hard on the speed of technology. It is that the accelerator on human institutions also needs to be pressed harder to govern increasingly powerful technology.

Is There a Third Way?

I have spent much of my professional life moving between science, industry, international diplomacy, universities and even rural communities. One lesson has remained constant: No technology operates in isolation. Every technology sits inside an economic system, a political system, an ecological system and, ultimately, a human value system. AI is no exception.

Nature, our oldest teacher, offers another lesson. Evolution does not simply reward the organism that grows fastest. It rewards systems capable of adaptation, diversity, resilience and balance. In short, systems that are sustainable. Perhaps that is the wisdom we need for AI.

We should neither blindly nor cunningly slow the technology nor recklessly accelerate it for monetary reasons. We should accelerate what strengthens humanity—and slow what threatens humanity.

The AI revolution has arrived at a crossroads where technology, geopolitics and climate change meet. One road leads towards a race for technological supremacy. Another towards a race for control. But there is a Third Way!

Accelerate AI where it enlarges human capability, protects the planet and strengthens peace. Slow it where machine speed threatens ecological stability, international security and inequality and endangers the Sustainable Development Goals. That is not anti-technology. It is pro-humanity. That's the Third Way.

(The author is a noted environmentalist, former Director of UNEP, Coordinating Lead Author, IPCC 2007 (Nobel Peace Prize laureate), IIT alumnus, and Founder of the Green TERRE Foundation, Pune. The views expressed are personal. He can be reached at shende.rajendra@gmail.com.)