Marcus—I worry that the intuition that ‘consciousness arises more easily from analogy than from digital processing’ it utterly unreliable, and depends on humans getting very confused about how to square their intuitive physics module with their intuitive psychology module.
Specifically, our intuitive physics module gives us the sense that ‘analog stuff’ is smooth, creamy, fine-grained, subtle, gentle, etc—and our intuitive psychology module says ‘Oh yes! That’s just like my smooth, creamy, fine-grained, subtle, gentle consciousness!’
By contrast, we have a stereotype of ‘digital stuff’ as brittle, jarring, binary, coarse-grained, simplistic, etc, and imagine that such a fine thing as human consciousness could never arise from such a substrate.
As someone who’s taken an evolutionary functional view of the human mind for many decades of research, I simply don’t trust these ‘philosophy of mind’ intuitions at all.
As some of the other comments here have said, a large deep learning network can be described in quite ‘analog’ terms, or in quite ‘digital’ terms, and neither is very informative about whether the network could actually give rise to sentient experience.
Marcus—I worry that the intuition that ‘consciousness arises more easily from analogy than from digital processing’ it utterly unreliable, and depends on humans getting very confused about how to square their intuitive physics module with their intuitive psychology module.
Specifically, our intuitive physics module gives us the sense that ‘analog stuff’ is smooth, creamy, fine-grained, subtle, gentle, etc—and our intuitive psychology module says ‘Oh yes! That’s just like my smooth, creamy, fine-grained, subtle, gentle consciousness!’
By contrast, we have a stereotype of ‘digital stuff’ as brittle, jarring, binary, coarse-grained, simplistic, etc, and imagine that such a fine thing as human consciousness could never arise from such a substrate.
As someone who’s taken an evolutionary functional view of the human mind for many decades of research, I simply don’t trust these ‘philosophy of mind’ intuitions at all.
As some of the other comments here have said, a large deep learning network can be described in quite ‘analog’ terms, or in quite ‘digital’ terms, and neither is very informative about whether the network could actually give rise to sentient experience.