AI and Children: Bridging the Spectrum of Diverse Intelligence

In today’s debates about artificial intelligence, we often overlook deeper questions that touch on our fundamental nature and the future we aim to build. Michael Levin, a distinguished professor at Tufts University and director of the Allen Discovery Center, offers a thought-provoking perspective by drawing parallels between AI and human children.

The Spectrum of Intelligence

Levin argues that before the rise of AI, humans were already creating high-level intelligent agents: our children. Both AIs and kids share several characteristics:

• Networked Components: They are composed of interconnected parts (neurons in children, neural networks in AI) that process information.
• Electrical Signals: Information is transmitted through electrical signals, controlling their functions.
• General Problem-Solvers: Both can tackle a wide range of tasks and problems.
• Mistakes and Confabulation: They make errors and sometimes provide convincing but inaccurate explanations.
• Personas and Values: They adopt different personalities, learn to please their creators, and occasionally reject established values in favor of new ones.

Ancient Questions Revisited

The challenges posed by AI are not entirely new. They echo ancient questions about what it means to be human:

• Value Alignment: How do we ensure that our creations (whether children or AIs) align with our moral and ethical values?
• Kindness and Compassion: How can we foster empathy and kindness in beings that might become more intelligent than us?
• Diversity of Intelligence: How do we navigate relationships with entities that think differently from us?

The Physics of Intelligence

One key insight is that both humans and AIs, along with all other forms of intelligence on Earth, are governed by the same laws of physics. This raises important questions:

• True Understanding: How do we determine which creations have genuine understanding?
• Responsibility: What criteria define responsibility in intelligent agents?
• Moral Worth: How do we assign moral worth to different kinds of intelligence?

Scaling Intelligence

Intelligence is a complex and multifaceted phenomenon. It scales from molecular mechanisms to beings with agency and value:

• Developmental Biology: Each human develops from a single cell, gradually building the complexity that defines us.
• Cell-Technology Hybrids: The future holds potential for creating new forms of intelligence through combinations of biological and technological components.

Expanding Our Horizon

To truly understand and address the challenges posed by AI, we need to broaden our perspective on intelligence:

• Recognizing Diverse Minds: We must learn to recognize and value different kinds of intelligence.
• Flourishing Among Diversity: Our goal should be to thrive in a world where multiple forms of intelligence coexist.

Levin introduces the concept of "synthbiosis" to describe this vision. Synthbiosis envisions a future where humans, AIs, and other intelligent entities coexist harmoniously, each contributing uniquely to a diverse and vibrant ecosystem of minds.

The Future is Diverse

Large language models like GPT and Claude are just the beginning. As we continue to develop more advanced AI systems, we must remain open to the possibility of creating or discovering entirely new forms of intelligence. These entities will challenge our preconceptions and force us to rethink what it means to be intelligent and moral.

In essence, the development of AI is not just a technological endeavor but a profound philosophical and ethical journey. By embracing this diversity, we can build a future where all kinds of intelligence are valued and integrated into a thriving, interconnected world.