Never. Code is software, software will never be sentient, it could only simulate it more or less accurately. Code is informational, like saying a cooking recipe could become sentient. Neurons are physical, hardware. Everything in the brain is hardware, or in this case, wetware.
But physical neural networks, as in neuromorphic computers? Now you're talking.
The hardware doesn't matter. You can be using Conway's game of life, computers, or the human brain to do the data processing and computation. It's called substrate independence. It's the same kind of thing in the end. If your computational device can do universal computation it is able to emulate computation on any other device. The brain does data processing and so does a computer, and there is no reason why the computer can't emulate the brain.
A computer is good at memory retrieval and computational speed, but can't deal with purely subjective data like the human brain can. It can only handle functions that are already built into it, and no one has ever produced even a conceptual design for one that can evolve unique functionalities.
Here's something you might find interesting. Have you heard of the ? It tells us that we can transform a continuous signal into a discrete signal if we sample at twice the frequency of the signal. Just because something is continuous doesn't mean it can't be represented digitally and the theorem tells us how to do that.
It's not the same thing because computers the way we conceive them do not process information the same way as brains do at all.
(post is archived)