The Space Of Possible Minds




They are assembled from components that are networked together to process information. Electrical signals propagate throughout, controlling every aspect of their functioning. Being general problem-solvers, many of them have high IQs, but they routinely make mistakes and confabulate. They take on different personas, learning to please their makers, but sometimes they abruptly turn on them, rejecting cherished values and developing new ones spontaneously. They convincingly describe things they don’t really understand. And they’re going to change everything.

I’m talking, of course, about our children.

Long before AI, we were creating high-level intelligent agents: kids. While the challenges that AIs provoke today seem novel, in reality, they echo fundamental and ancient questions about what it means to be human. How can we make sure new generations of beings align with our values? How do we ensure that our creations treat themselves and us with kindness and compassion? How do we calibrate relationships with those who are not like us? What happens to us, and to humanity, if each generation becomes smarter and more adventurous than their parents? And given that humans, AIs and every other form of intelligence on Earth obey the same laws of physics, how exactly do we determine which of our creations have true understanding, responsibility and moral worth?

Answering these questions and understanding the true challenges posed by the development of today’s AIs requires appreciating just how diverse intelligence is. After all, each of us developed slowly and gradually from a single cell, and we are only now beginning to understand how intelligence scales from molecular mechanisms into beings with agency and value. What other bodies and minds can be made by cells and cell-technology hybrids? Large language models like GPT and Claude are just the beginning.

Our goal must be to expand our narrow ability to recognize minds and learn to flourish among many different kinds of intelligence, from humans to AI to weirder and more wonderful beings that will arrive in the future. You might call this concept synthbiosis: the quest to develop mutually beneficial relationships between radically different beings that are evolving together, at different rates and in different directions, in the web of life on Earth and beyond.

The Space Of Possible Minds

Humans and today’s AIs are just two data points on a spectrum of intelligent beings, which is only going to become vaster and more complex. “Diverse intelligence” encompasses a wide variety of unconventional beings that already exist or could evolve beyond the familiar materials, forms and functions with which the meanderings of mutation and selection have left them.

Like children, AIs offer humanity an incredible gift: They push us to explore ideas of how we embody the true understanding and agency we believe we have, what really matters to us as individuals and societies, and where we want to go as a species. The forthcoming diversity of beings, and our expansion into the enormous range of possible embodiments of sentience, will shatter untenable old narratives of what we are, what it means to change, what we can become and what we should value. This requires us to rapidly mature our assumptions about intelligence, agency, cognition and life itself. We must shift the conversation around AIs from “What can they do?” to “With such a wide range of beings that deserve moral consideration, how can we care for each other?”

Note that I am not arguing that currently popular AI architectures have anything like a human mind or that today’s creations exploit the key principles and self-construction processes needed for agency and selfhood as seen throughout the biological world. Nor am I arguing that AI doesn’t raise a few unique problems.

But the recent emergence of sophisticated AIs does confront humanity with the opportunity to shed the stale categories of natural and artificial — to define what we want a mature human species to be. Focusing on the current technology and what it might do to us distracts from much more interesting and important questions about us and what kind of place we want to occupy in the future, all of which must be resolved for our flourishing as a mature species.

“The forthcoming diversity of beings will shatter untenable old narratives of what we are, what it means to change, what we can become and what we should value.”

Terms such as life, machine, mind, sentience or robot have never been crisp categories objectively describing a living or artificial system. Instead, we should think of them as relationships — ways for each system to relate to others, all of which carry strong implications for the utility and ethics of the resulting interactions.

Already today, the continuum of life is sprinkled with cyborgs and other hybrids of biological and technological components, like people with embedded insulin pumps or cochlear implants to enable hearing. Other modifications will include enhancements such as entirely new cognitive modes via virtual reality or brain implants that can better control limbs or organs. Mind-machine interfaces will radically expand the capabilities and borders of our embodied selves. Large language models too are chimeras — trained on human creative output to imitate people.

Seguir leyendo: Noéma

Imagen de Gordon Johnson en Pixabay