Why AI cannot be conscious
This essay is based on a report I wrote for my writing class back at university. It’s even more relevant today in an age where investors genuinely think LLMs understand them and startups want to help upload your mind. Although no background in philosophy is needed, the writing style is more academic and assertive than usual because that’s what the course wanted from me so I apologize in advance (at least my instructor said it was the best paper :p)
Background
The belief that sophisticated software systems can replicate consciousness is an implication of materialist metaphysics. Materialism is the philosophical view that all phenomena, including mental phenomena such as consciousness, can be reduced to matter. In this view, even seemingly immaterial phenomena such as emotions or the perception of colors can be reduced to physical processes in the brain. In other words, they can be simulated on a computer. This means that the existence of conscious AI is highly reliant on materialist claims. Thus, AI can not be conscious if materialism is false. Yes, I’m tricking you into reading a philosophical essay by using a tech title.
A clear distinction must be made between the mind and the brain in order to examine materialism properly. The brain is a biological organ consisting of billions of neurons, while the mind is the set of all mental perceptions that one experiences. Although there is a strong correlation between the mind and the brain, they are not equivalent. For example, the neural activity associated with observing a color differs from the subjective experience of seeing the color itself. The latter can not be communicated symbolically or detected using an instrument. A blind scientist who studies the neuronal activity of colors will not be able to experience them directly, regardless of how much they know about their associated neurons. Materialism reduces the mind to the brain. That is, it claims that the mind is a by-product of brain activity and that consciousness can be generated through the stimulation of specific neurons. Or simply put, the mind is a kind of computer.
Is materialism even true?
Attempting to create consciousness becomes an obvious and exciting step towards the advancement of mankind. However, before spending an enormous amount of resources on such projects, I encourage you to examine the coherence of materialism itself. It’s true that materialism is the most commonly held position among academics. However, its popularity doesn’t necessarily make it true (that’d be an appeal to popularity). In fact, the dominance of materialist metaphysics is often attributed to historical accidents and momentum, not epistemic superiority.
Although most natural phenomena can be explained precisely and accurately in terms of matter interactions, materialism has yet to provide a direct and concrete explanation of conscious experience. No one-to-one relationships that map mental perceptions with the activation of certain neurons have been discovered. In fact, neuronal participation is highly context-dependent: a neuron sometimes contributes to an experience, but sometimes it does not. Under materialism, knowing the physical state of neurons should be sufficient to predict their corresponding subjective experience. Although many correlations have been discovered, the fact that neuroscientists have not been able to find a mind-brain mapping despite decades of research is troublesome. One could argue that future breakthroughs will improve our understanding of the mind in a way that is consistent with materialism, but that’s a speculative claim that relies on an open-ended problem in science. It’s not rational to insist that subjective experience is fully physical given the lack of explanatory account of consciousness.
What’s even more challenging for materialism is the fact that subjective experience is categorically different from matter. The neural activity associated with observing the color red is fundamentally distinct from the subjective experience of red itself. Reductive materialism states that one produces the other without providing an ontological mechanism that bridges them, not even in principle. Notice how this is not an issue of empirical evidence nor scientific progress, but an issue within materialism itself. This is the the hard problem of consciousness.
Adjacent views
Many attempts have been made to resolve the hard problem of consciousness without changing materialism too much. Panpsychists, for instance, claim that conscious experience is a fundamental property of the universe, present within all things. While this eliminates the “gap” between the material and the mental, it’s difficult to imagine that everything, including everyday objects or subatomic particles, could have a small degree of consciousness. There’s simply no empirical evidence of such a thing. Furthermore, it doesn’t give an account for how smaller units of consciousness combine to produce complex conscious lifeforms like humans. It’s a fascinating proposition, but the lack of evidence severely undermines its plausibility.
Another attempt is to go as far as to deny the existence of conscious experience, suggesting that it’s an illusion generated by the brain… As if we’re all secretly p-zombies in denial. I find such claims amusingly absurd. Mental experience is the only thing that is directly and undoubtedly real. All knowledge is built on the assumption that perceptions are grounded in reality. It’s much more reasonable to admit that materialism is inadequate than undermining your own mind.
Conclusion
Simulating the brain or pushing the boundaries of AI are fantastic projects and I’d love to help you with that. All I ask of people is some epistemic humility in their endeavours. Indeed, once we admit what we don’t know, we can approach such efforts with open-minded curiosity, rather than ideological dogma that has no place in the pursuit of knowledge.