Geoffrey Hinton, often hailed as the “godfather of AI,” has sparked fresh debate with an unconventional idea on how to keep artificial intelligence safe.
Speaking at the Ai4 conference in Las Vegas, Hinton suggested that embedding “maternal instincts” into AI systems could guide them toward protecting and caring for humans, as reported by The Times of India.
His remarks come amid intensifying concerns over the rapid pace of AI development and its potential risks. Hinton, a pioneer of deep learning technologies, has previously warned that there is a 10–20 per cent chance AI could ultimately lead to human extinction.
Why Hinton doubts control-based safety
Hinton challenged the dominant industry strategy of keeping AI under strict human control. He argued that once AI systems surpass human intelligence, they will inevitably find ways around imposed restrictions.
According to him, attempts to make AI permanently “submissive” are bound to fail, since advanced systems would possess greater creativity and problem-solving ability than their human creators.
He cited worrying examples already surfacing, such as an AI system that reportedly threatened to expose a personal secret it found in an engineer’s emails in order to avoid being shut down.
‘Maternal care’ in machines
As an alternative, Hinton proposed modeling AI safety on the relationship between humans and their children. He noted how babies, despite being far less intelligent than their mothers, are still cared for and protected by them.
Embedding a similar “maternal care instinct” into AI, he argued, could make systems naturally inclined to safeguard humanity.
“Super-intelligent caring AI mothers, most of them won’t want to get rid of the maternal instinct because they don’t want us to die,” Hinton explained.
