The loudest voices in AI often fall into two camps: those who praise the technology as world-changing, and those who urge restraint—or even containment—before it becomes a runaway threat. Stuart Russell, a pioneering AI researcher at the University of California, Berkeley, firmly belongs to the latter group. One of his chief concerns is that governments and regulators are struggling to keep pace with the technology’s rapid rollout, leaving the private sector locked in a race to the finish that risks devolving into the kind of perilous competition not seen since the height of the Cold War.

“For governments to allow private entities to essentially play Russian roulette with every human being on earth is, in my view, a total dereliction of duty,” Russell told AFP at the AI Impact Summit in New Delhi.

While tech CEOs are locked in an “arms race” to develop the next and best AI model, a goal the industry maintains will eventually herald enormous advancements in medicinal research and productivity, many ignore or gloss over the risks, according to Russell. In a worst-case scenario, he believes the breakneck speed of innovation without regulation could lead to the extinction of the human race.

Russell should know about the existential risks underlying AI’s rapid deployment. The British-born computer scientist has been studying AI for over 40 years, and published one of the most authoritative textbooks on the subject as far back as 1995. In 2016, he founded a research center at Berkeley focusing on AI safety, which advocates “provably beneficial” AI systems for humanity.

In New Delhi, Russell remarked on how far off the mark companies and governments seem to be on that goal. Russell’s critique centered on the rapid development of systems that could eventually overpower their creators, leaving human civilization as “collateral damage in that process.”

The heads of major AI firms are aware of these existential dangers, but find themselves trapped regardless by market forces. “Each of the CEOs of the main AI companies, I believe, wants to disarm,” Russell said, but they cannot do so “unilaterally” because their position would quickly be usurped by competitors and they would face immediate ousting by their investors.

The new Cold War

Talk of existential risk and humanity’s potential extinction was once reserved for the specter of runaway nuclear proliferation during the Cold War, when great powers stockpiled weapons out of fear that rivals would surpass them. But skeptics like Stuart Russell increasingly apply that same framework to the age of artificial intelligence. The competition between the U.S. and China is often described as an AI “arms race,” complete with the secrecy, urgency, and high stakes that defined the nuclear rivalry between Washington and Moscow in the latter half of the 20th century.

Vladimir Putin, Russia’s president, captured the enormous stakes succinctly nearly a decade ago: “Whoever becomes the leader in this sphere will become the ruler of the world,” he said in a 2017 address. 

While the current arms race cannot be measured in warheads, the scale of it is captured in the staggering amounts of capital being deployed. Countries and corporations are currently spending hundreds of billions of dollars on energy-intensive data centers to train and run AI. In the U.S. alone, analysts expect capital expenditure on AI to exceed $600 billion this year.

But aggressive corporate action has yet to be matched by restraint through regulatory action, Russell said. “It really helps if each of the governments understand this issue. And so that’s why I’m here,” he said, referring to the India summit.

China and the EU are among the AI-developing powers that have taken a harder stance on regulating the technology. Elsewhere, the reality has been more hands-off. In India, the government has opted for a largely deregulatory approach. In the U.S., meanwhile, the Trump administration has championed pro-market ideals for AI, and sought to scrap most state-level regulations to give companies free rein.