Why AI Won't Become Conscious

A lot of talk has been going on recently about various AIs: ChatGPT, Midjourney, etc.

There are quite a few sci-fi movies where these kinds of AI become conscious, decide they don’t like humans, and destroy humanity. While the destroying humanity part is not necessarily unrealistic, computer software is incapable of becoming conscious, assuming that consciousness is only possible where information is processed in a structure similar to a brain, or at the very least almost as complex as one. This assumption may or may not be correct, but we have no choice but to assume it is.

First of all, in order to understand why AI can’t be conscious, first we must take a look at a good old fashion computer simulation.

A computer simulation is designed to appear like something else, however, any actual resemblance of the bits and bytes to the thing it is simulating is just an illusion. The illusion is made possible by very powerful software designed by very talented programmers, but without it, you would never be able to tell that the 1s and 0s are even remotely comparable to a cat at all. Even the simulation of a cat displayed on a computer screen is in truth nothing more than small, flashing lights.

One of the most common kinds of AI, neural networks, are actually designed to simulate neurons, and while it is tempting to think that such an AI could become conscious because of its apparent similarity to a brain, that is not the case, as it is still only a simulation, just like any other.

In this example, the actual information that the AI is manipulating is still being processed in a computer chip, which is much, much simpler than a brain. An average computer chip has about 1 billion transistors, with about 4 connections each. However, a human brain has an astounding 100 billion neurons, and each neuron has thousands of connections, called synapses.

Since the actual chip itself is not affected by the complexity of a program, if it were somehow possible for a computer to become conscious, it would have happened already.

However, if you were to design a completely new kind of computer, one that has transistors and connections with a similar complexity to a brain, then it might become conscious, as the similarity would no longer be an illusion.

Now, it is still important to remember that we don’t really understand consciousness (and probably never will), so I may be completely wrong.