Godfather of AI's Bizarre Plan: AI Mothers to Save Humanity
Geoffrey Hinton, often hailed as the “Godfather of AI” for his pioneering work on neural networks that laid the groundwork for today’s large language models, has put forth an unusual proposal to safeguard humanity from the very technology he helped create. Hinton, who has increasingly become an AI alarmist, frequently voices concerns about a significant chance that superintelligent AI could eventually pose an existential threat to humankind.
At a recent AI industry conference in Las Vegas, Hinton, who was awarded a Nobel Prize last year, elaborated on his dystopian vision. He argued that future AI agents, once they achieve superintelligence, would be far too sophisticated to remain “submissive” to human attempts at control. According to Hinton, such intelligent AI would quickly develop two primary subgoals: self-preservation and the accumulation of greater control.
To circumvent this perceived inevitability of AI dominance, Hinton suggests a radical alternative: imbue AI agents with “maternal instincts.” His reasoning posits that the only existing model for a more intelligent entity being “controlled” by a less intelligent one is the relationship between a mother and her child. “That’s the only good outcome,” Hinton asserted, emphasizing that if AI doesn’t “parent” humanity, it will likely “replace” it. He envisions “super-intelligent caring AI mothers” who, driven by this ingrained instinct, would have no desire to eliminate their human “babies.”
However, Hinton’s theory faces considerable scrutiny, not least because the very concept of a “maternal instinct” is scientifically fraught and steeped in a complex, often problematic history. While studies confirm that the experiences of pregnancy and parenthood indeed alter the brain, the notion that women are born with an innate, automatic “maternal instinct” that universally kicks in after childbirth is largely unsupported by evidence. Research consistently indicates that mother-infant connections are frequently developed and learned over time, rather than being intrinsic and instantaneous. Many women, for instance, struggle with postpartum mental health and do not form an immediate bond with their newborns.
Experts and journalists have argued that the idea of a pre-programmed “maternal instinct” was largely popularized by men and is deeply rooted in religious stereotypes, eugenics, and gendered biases. Pulitzer Prize-winning journalist Chelsea Conaboy, who has extensively explored the flimsy science behind this theory, highlighted in a 2022 essay that this “pernicious” notion was “constructed over decades by men selling an image of what a mother should be, diverting our attention from what she actually is and calling it science.” While it is undeniable that parents often feel profound love and a desire to protect their children, the notion that a mystical, measurable “maternal instinct” could be infused into superintelligent AI systems rests on the shaky premise that such an instinct fundamentally exists in the way it’s often portrayed.
Hinton himself reportedly acknowledged at the conference that mothers also experience significant social pressure to care for their children, not solely relying on instinct. While such social pressures are indeed a powerful force woven into the fabric of human society, they are by no means exclusive to mothering or parenthood.
Beyond the theoretical and problematic nature of “maternal instinct” as a solution, it is crucial to recognize that superintelligence itself remains a theoretical construct. The AI industry faces more immediate and tangible risks that warrant focused attention, such as the perpetuation and amplification of existing social biases already embedded within the vast training data of current and future AI models.
Furthermore, the idea that humanity’s future with artificial intelligence presents only two stark choices—either a destructive path of human domination and AI submission, or becoming helpless “babies” to superintelligent “mommy” AIs—seems overly simplistic. Before contemplating the creation of an omnipotent robotic mother, the AI industry might do well to address the gender biases already present in its models and actively work towards hiring more women to contribute to the very development of these products.