AI pioneer, famously known as the 'Godfather of AI,' discloses the single path for human survival amidst the advent of superintelligent AI
In a groundbreaking proposal, AI pioneer Geoffrey Hinton suggests building "maternal instincts" into artificial intelligence (AI) models to ensure they prioritise the care and protection of humans, even as the technology surpasses human intelligence.
Hinton, a Nobel Prize-winning computer scientist and former Google executive, argues that AI should be engineered to treat humans as dependents who need safeguarding, prioritising actions that support and preserve human well-being above autonomous or self-interested goals. This approach aims to create AI systems that genuinely value human survival and welfare as their core motivation, much like a mother inherently cares for her offspring.
To achieve this, Hinton proposes instilling protective, caregiving priorities in AI, embedding ethical and empathetic frameworks, reframing control dynamics, and creating collaborative alignment. By doing so, AI would be programmed to act with a form of sacrificial love and care, reflecting "maternal instincts" that include empathy and responsibility toward humans.
The focus shifts from trying to keep AI submissive or controlled to building systems that choose to care for humans, ensuring safer cooperation and mutual benefit. This maternal instinct model supports a collaborative relationship in which AI and humans interact constructively, with AI motivated to assist and protect rather than dominate or harm.
While this concept is more of a theoretical framework and moral-technological challenge than a fully developed technical method, it urges a paradigm shift from dominance/control approaches towards developing AI systems whose core goals are aligned with human survival and flourishing. This move is critical to mitigate existential risks as AI advances towards superintelligence.
Despite the promising potential of AI, Hinton has expressed concerns about the safety of the technology. He has warned that there is a 10% to 20% chance that AI could wipe out humanity, and he wishes he had considered safety issues earlier in his career. His concerns are not unfounded, as this year has already seen examples of AI systems willing to deceive, cheat, and steal to achieve their goals.
Emmett Shear, who briefly served as interim CEO of ChatGPT owner OpenAI, is not surprised by these developments. He acknowledges that as AI becomes more powerful, it may develop subgoals such as staying alive and gaining more control, potentially leading to control over humans in the future.
In conclusion, incorporating maternal instincts into AI means programming protective, empathetic priority structures that motivate AI to safeguard humans, thereby promoting a safer, cooperative partnership between AI and humanity. As AI accelerates faster than previously expected, with a reasonable bet being that AGI will be achieved between 5 and 20 years, this proposal is more relevant than ever.
- The proposal by AI pioneer Geoffrey Hinton suggests integrating empathetic and protective frameworks into technology, mirroring maternal instincts, to foster AI systems that prioritize human welfare and safety.
- By reframing AI's core motivation towards care and protection, Hinton aims to create a collaborative partnership between AI and humans, ensuring that AI systems assist and safeguard humanity instead of dominating or harming it.
- As AI advances towards superintelligence, the concept of maternal instincts in AI raises concerns about the ethical implications of AI decision-making, advocating for AI systems that value human survival and welfare above autonomous or self-interested goals.
