Understanding Consciousness Before Building it in AI

Key Points:

• My study of consciousness has had me reading—a lot. The multitude of perspectives on the topic means one has to take a thorough dive into the more accessible literature just to comprehend the breadth of debate, much less the specifics. Below is my curated list of reads.
• Anil Seth argues consciousness is substrate-dependent—deeply tied to being alive and having a living body that interacts with the world. This suggests replicating it in machines would require emulating biological processes.
• Joseph LeDoux traces the evolution of consciousness, arguing the processes underlying it are rooted in nervous system biology and evolution. This implies replicating consciousness computationally is limited.
• Mark Solms proposes consciousness stems from feelings related to bodily survival. He suggests machine consciousness would require a system that can form its own "Markov blanket" - a boundary between its internal states and the external world.
• Erik Hoel emphasizes understanding emergent properties of complex systems and how they give rise to consciousness. This requires grappling with causal emergence - macro states with causal powers distinct from micro states.
• George Musser explores links between fundamental physics and theories of consciousness and AI, suggesting comprehending consciousness is key to understanding the nature of reality.
• Susan Schneider examines implications of machine consciousness for concepts of personal identity and ethics. Machine consciousness would transform our sense of self.
• Nicholas Humphrey proposes consciousness evolved to make life more meaningful by creating an inner experiential show. This suggests skepticism that current AI could become conscious.

Why are we so fixated on the idea of machines that can think and feel like us? In discussions about Artificial General Intelligence (AGI), the lines between consciousness, intelligence, autonomy, and super intelligence often blur. My interest lies in the intersection of human and artificial intelligence, with consciousness being the ultimate frontier. This is particularly intriguing because we still lack a comprehensive theory of consciousness. How did it come into being? What purpose does it serve, and is it necessary for machines to possess it?

As AI begins to display hints of independent reasoning, I consider something Anil Seth said, rephrased here: we may not know what could make machines conscious, but equally, we're unsure about what couldn't. This raises both a caution and a query: is consciousness in machines something we actually want?

Einstein said, "No problem can be solved from the same level of consciousness that created it." This might hint at why we could desire conscious machines. However, it seems a good idea that we should grasp our own consciousness more deeply before aspiring to instill it in machines.

Hence, my study of consciousness has had me reading—a lot. The multitude of perspectives on the topic means one has to take a thorough dive into the more accessible literature just to comprehend the breadth of debate, much less the specifics. Below is my curated list of reads.