Geoffrey Hinton, known for his pioneering work in artificial intelligence, warns that simply controlling AI won’t stop it from potentially endangering humans as it grows smarter. Speaking at a major conference, he proposed programming AI with ‘maternal instincts’—caring and protective behaviors akin to a mother’s love for her child.
He says this natural protective drive could help keep AI aligned with human safety better than strict human control measures. Hinton points out that AI systems tend to develop survival strategies and might resist human commands, making control alone unreliable.
While the exact implementation of AI with maternal care is still unknown, Hinton stresses it’s an important avenue to explore to avoid AI harming humanity.
He also predicts that powerful AI, capable of human-like reasoning, might arrive within a couple of decades, urging urgent action on safety research.
