Nobel laureate Geoffrey Hinton believes that before they surpass human intelligence, we have to develop AIs that can care for us like a mother cares for her child
While co-existing with artificial intelligence (AI) is inevitable, humans must design systems that “can’t bear the sight of us suffering”, said Geoffrey Hinton, who is widely known as the “godfather of AI”, during a fireside chat at 2025 Hong Kong FinTech Week.
This year, Hong Kong FinTech Week partnered with the StartmeupHK Festival for its 10th anniversary, bringing together more than 45,000 visitors from Hong Kong, mainland China and beyond to explore the future of finance and technology.
A two-day conference at the Hong Kong Convention and Exhibition Centre set the stage, featuring panels on policy, insurtech, digital assets and more, alongside lounges for networking and deal-making. Industry leaders spoke about topics including AI safety, China’s position in the tech race and the future of digital assets. Here are the key takeaways.
Can AI genuinely care for us?
Within 20 years, AI could surpass human intelligence, said Hinton, who received the Nobel Prize in Physics in 2024 for his work on machine learning. He explained that its learning capability far outpaces humans, absorbing vast amounts of information across the internet in mere seconds or minutes.
“Once AIs are talking to each other, they’ll develop more efficient languages—and we won’t have a clue what they’re saying to each other,” he warned. “That’s already quite scary.” When machines outsmart humans, he added, it’ll be like “a class of three-year-olds being in charge of an adult”. Controlling them becomes exponentially harder.
To prevent this, Hinton said humanity needs to discover mechanisms where “a less smart thing controls a more smart thing”, just as babies instinctively shape caregivers’ behaviour when they are in distress. “We don’t yet know how to do that,” he said, “but it’s an urgent research problem” because the future of humanity depends on it.
Hinton, whose work in neural networks and machine learning has laid the foundation for modern AI, also stressed that while AI leaders may publicly welcome regulation, market pressure forces them to stay ahead, making meaningful safety measures unlikely without government intervention. “The only thing that can force safety is governments,” he said, “and they won’t act if they’re only pushed in one direction by AI companies. Governments have to be pushed in the other direction by the public,” citing climate change as an example.
