Geoffrey Hinton, known as the “Godfather of AI,” fears that superintelligent AI will challenge human dominance—but he has a proposal that could reframe AI assistants as AI mothers.
Article contents:
- Introduction: The Godfather of AI’s fear
- The “maternal instinct” solution
- “The only good outcome”
- The timeline for advanced artificial general intelligence
- Previous warnings: From extinction to unemployment
- Who is at risk? Desk jobs first
- An unsolved technical challenge
- Conclusion: Can a mother’s instinct save us?
Introduction: The Godfather of AI’s fear
In a keynote address at the Ai4 conference in Las Vegas, 78-year-old Geoffrey Hinton predicted a future where AI could assert control over humans as easily as an adult interacts with a 3-year-old child, getting them to complete a task with the promise of candy. Hinton said: “The AI will be much smarter than us.”
Hinton is known as the “Godfather of AI” due to his pioneering studies that laid the foundation for current AI systems, such as ChatGPT and other chatbots. Hinton began his work in the late 1970s and eventually won the 2024 Nobel Prize in Physics for his work.
The “maternal instinct” solution
In his speech, Hinton suggested training AI to have “maternal instincts” so that it is programmed to care deeply about people. This way, advanced AI systems would be trained with the same instincts possessed by a mother looking out for the survival of her children. He stressed that “the only model” of a more intelligent being controlled by a less intelligent being is “a mother controlled by her child.”
“The only good outcome”
Hinton told CNN Business: “This is the only good outcome. If it’s not going to take care of me like a parent, it’s going to replace me.” This strong statement reflects how seriously Hinton views the existential risks posed by superintelligent AI. He sees no middle ground; either AI cares for us like a parent, or it will surpass us.
The timeline for advanced artificial general intelligence
Hinton also shortened his expected timeline for advanced artificial general intelligence (AGI), or AI that exceeds human intelligence. Instead of predicting it could take 30 to 50 years before AGI emerges, Hinton said a “more reasonable bet” is five to 20 years. This acceleration in the timeline increases the urgency of finding solutions to the control problem.
Previous warnings: From extinction to unemployment
Hinton has spoken about the impact of AI on humanity before, and his warnings have ranged from extinction to mass unemployment. For example, in December, Hinton predicted that there is at least a 10% chance that AI will wipe out humanity and lead to human extinction within the next 30 years.
Who is at risk? Desk jobs first
In a podcast appearance in June, Hinton predicted that AI will replace everyone in desk jobs, noting that professions such as paralegals and call center representatives are most at risk. He said it will take “a long time” before AI takes over physical tasks and manual labor, making these professions the least at risk for now.
An unsolved technical challenge
Hinton acknowledged that he does not know how to technically achieve the task of creating an AI with maternal instincts, but he stressed that it is crucial for AI researchers and developers to work toward it. This represents a formidable challenge for the AI community, as programming complex and undefined concepts like “care” and “maternal love” into a computational system remains out of reach.
Conclusion: Can a mother’s instinct save us?
Geoffrey Hinton’s thought-provoking proposal offers a fresh perspective on the AI safety problem. Instead of focusing solely on restrictions and rules, he suggests building a fundamental positive drive into the heart of AI. It remains to be seen whether this approach is technically feasible, but it certainly raises a profound question: Could our most powerful biological relationship—that of a mother and her child—be the key to ensuring our survival in the age of superintelligent machines?
