Artificial intelligence (AI) is changing the world we live in, from classrooms to offices and now battlefields. Countries such as the US, China and Russia are competing to integrate AI into the military. Recently, when Ukraine used Ai-Enhanced drones against Russia, they even saw the power of machine learning on the battlefield. These drones were able to autonomously navigate and strike targets, controlled from thousands of miles away. While military AI is making modern warfare even more sophisticated, AI godfather Geoffrey Hinton believes that pushing the military to AI will make it easier for wars to begin and fights safe.
In a recent interview with journalist Katie Courick, Hinton warned that a deadly autonomous weapon, often referred to as killer robots, is one of the most dangerous applications of AI technology. “Deadly autonomous weapons, those that decide who to kill, who to kill yourself, are a great advantage if a wealthy country wants to invade a poor country,” he said.
Hinton explained that the danger of deploying advanced AI on the battlefield lies in reducing human and political conflict costs. “It's that their citizens will return to body bags that will stop the rich country that invade poor countries,” he said. “If there are deadly autonomous weapons, instead of dead people coming back, dead robots will come back.
He also emphasized that the country is already using AI to restructure wars. Drones costing hundreds of dollars can destroy millions of dollars of tanks, and traditional fighters with human pilots are beginning to look outdated. “Fighters with people now are stupid ideas. If we can put AI in, AIS can withstand much greater acceleration and we don't have to worry too much about the loss of life,” he said.
His comments, reflecting what Elon Musk said a few months ago, arguing that drones are the future of war, rather than expensive fighter and bomber jets that now serve as military main offensive power. “Chinese drones like DJI and military models cost a thousandth of the F-35, but they could destroy one in a few seconds,” Musk recently said.
According to Hinton, the use of advanced technology in wars does not reduce casualties and may instead promote military adventures. He points out that while soldiers' lives have traditionally been a limiting factor for governments, machines that drive AI can remove their deterrent and can encourage weapon manufacturers to introduce powerful weapons. “That's great because it's expensive to replace for a military industrial complex,” he said.
AI vs. humans face a dangerous future
The risks on the battlefield of AI are amazing, but Hinton also spoke about other long-term risks. He emphasizes that most experts now believe that AI can outweigh human intelligence within five to 20 years, and that when machines become out of control, pose existential risks. “We know there are few examples of smarter things controlled by smarter things,” he warned.
Hinton also noted that short-term AI threats are already clear, from machine learning-powered cyberattacks to creating dangerous viruses and over-realistic fake videos.
And of course, unemployment is one of the big fears about AI. Even Hinton believes that AI can destabilize the economy by driving away workers. He predicted that call center roles, paralegal research, and routine programming would be one of the first jobs at risk, and that more complex or creative roles could last longer. “When one person can do five jobs, that should mean there's more for everyone,” he said. “But in the systems we have, the owners who implement AI will become richer and the unemployed will become poorer.”
In particular, this is not the first time Hinton has spoken about the dangers of AI. After leaving Google in 2023, he has spoken a lot about the potential risks that AI is not under control and has repeatedly called for global cooperation. He also explains how to use metaphors to handle AI. At one point, he compared AI to Tiger Cub, and while he wasn't threatened when he was little, he was destined to grow into something dangerous. He also suggested that to stop AI from hurting humans, developers need to train it with “mother's instincts.”
– end
