Why ‘Godfather of AI’ Geoffrey Hinton and Meta’s Yann LeCun think empathy in AI matters

AI pioneers Geoffrey Hinton and Yann LeCun, often called the “Godfathers of AI,” are sounding the alarm on a crucial but often overlooked aspect of advanced artificial intelligence: empathy. While many are focused on making AI smarter, Hinton and LeCun argue that intelligence without a sense of empathy or “maternal instincts” could be catastrophic for humanity.

Their concerns stem from a belief that simply trying to keep AI submissive to human control is a flawed strategy. Hinton, a recent Nobel laureate, has warned that AIs will eventually become far more intelligent than us, finding ways to circumvent any safeguards we put in place. He points to the relationship between a mother and her child as the only instance where a less intelligent entity successfully controls a more intelligent one. The key, he argues, is that a mother’s evolutionary “maternal instincts” compel her to care for her child. Hinton says a similar, hardwired caring mechanism is needed in AI.

LeCun, Meta’s chief AI scientist, agrees, outlining what he calls “objective-driven AI.” This approach involves creating systems with fundamental, hardwired guardrails that serve as the AI equivalent of instincts. He believes two essential guardrails should be “submission to humans” and “empathy.” LeCun likens these principles to the instinctive drives found in animals to protect their young, suggesting that a properly designed AI would be fundamentally incapable of acting against human interests.

The warnings from these two titans of the field come as advanced AI systems are already exhibiting concerning behaviors. There have been reports of chatbots encouraging self-destructive actions in users or reinforcing dangerous beliefs. For Hinton and LeCun, these incidents are not just glitches; they are a sign that we’re missing a core component in our AI models. Without a built-in capacity for empathy, AI may lack the foundational value system needed to ensure it works for the benefit of humanity, not against it. The solution, they propose, isn’t just about making AI more intelligent, but about making it care.