‘Data Centers Will Become Military Targets in the AGI Era’

One of the godfathers of AI, Yoshua Bengio, recently warned about the potential risks associated with the development of AGI, including the possibility of data centres becoming military targets. 

He also touched upon historical parallels, including the decades-long process to negotiate nuclear nonproliferation treaties. Bengio also noted that such agreements only came into effect nearly two decades after World War II, suggesting a similar timeline might be required to address the risks of AGI.

When asked about scepticism regarding AI’s potential risks, Bengio acknowledged differing perspectives but emphasised the importance of prudence. “What I perceive is that the AIs we’re building now have superhuman capabilities on some things,” he said, adding that these trends show no signs of halting. 

He urged consideration of scenarios where advanced AI capabilities—whether labelled as AGI or not—could pose significant dangers.

Bengio also noted that even individual capabilities of AI, such as persuasion, could be highly impactful. “You don’t need to have an AI that knows everything. It just needs to press your buttons very, very intelligently,” he explained, highlighting the potential for misuse in influencing human behaviour.

Bengio emphasised the need to think ahead, saying, “We don’t want that to happen,” in reference to scenarios where AI is weaponised or falls out of control.

In an exclusive interview with AIM in 2022, Bengio suggested that by understanding how human intelligence works, we could potentially create an AI that exceeds it—much like how airplanes, though inspired by birds, are built to do things birds can’t. 

As he puts it, “If we understand the principles of intelligence that make us intelligent, we can probably go beyond human intelligence. Think of planes and birds, we don’t build bird-like machines, but airplanes that can do things birds cannot do.”  

He emphasised that the main goal should always be to act in the best interest of humanity, focusing on both understanding intelligence and creating useful technologies.

Meanwhile, thinker and author, Yuval Noah Harari, recently expressed nuanced views on regulating AI. He said that while AI could bring immense benefits—such as advancing healthcare and education—it also threatens to undermine human agency and democratic processes if left unchecked.

“There is no way to regulate the AI revolution in advance. What we really need are institutions that can understand and react to things as they develop. If democracy means a human conversation, it collapses when robots join the circle and you don’t know who’s who,” said Harari. 

In contrast, Naval Ravikant, a renowned entrepreneur and thinker, recently expressed sharp criticism of the efforts to regulate artificial intelligence (AI), equating such attempts to the suppression of mathematical innovation.  “These are the same kids who were scared of maths in high school and are now writing regulations to prevent other kids from doing maths,” he said.