From its continual appearance in sci-fi movies to Elon Musk declaring it “the most serious threat to the survival of the human race”, the idea of “artificial sentience” has become a major part of popular culture today. With this said, and after Elons warning, should we all be nervous for the creation of sentient AI in the near future, or is it just another celebrity trying to create a headline.
Today, artificial intelligence has taken over the world, whether it be in captcha, or facial recognition, artificial neural networks (a pattern recognition technique which mimics the brain in structure) are being used throughout our daily life. In fact, artificial intelligence has managed to exceed our own intelligence in multiple different areas. The most famous example of this would be in 1997 when IBM created the “Deep Blue” computer which managed to beat the reigning chess champion Garry Kasparov. A more modern example would be the use of convolutional neural networks to detect and predict the pathology of tumours in images. Although these two examples are highly intelligent, they are both incredibly specific, neither model can perform functions outside of their domain, therefore both models have very low, if any, general intelligence. However, even if they did possess general intelligence this does not equate to artificial consciousness.
To determine whether we have created artificial consciousness we must first define what it is to be conscious. The issue with this question is that it is primarily philosophical and therefore impossible to come to a conclusive answer. According to the school of thought of panpsychism, every object in existence, even individual atoms, experience some level of consciousness, albeit to varying degrees. Therefore, we could not strive for artificial consciousness as the programs we have created were already conscious. When we say artificial consciousness, we are instead referring to a comparable level of consciousness to humans, where we and our programs experience reality to a similar degree. This has been coined using the term “strong artificial intelligence”.
In the 1940s, two mathematicians made the proposition that the brain itself was simply a computer and could be replicated by a Turing machine. This arose to the philosophical school of thought known as the computational theory of mind, which later become central to the development of cognitive science. However, since then this school of thought has been highly disputed, with many people dismissing the idea that the brain is a highly sophisticated computer. With arguments such as Gödel’s incompleteness theorem, or the fact that a computer works with discrete data whereas the brain uses continuous and discrete processes. If correct, these objections point towards the possibility that strong artificial intelligence is impossible.
If we continue under the assumption that the brain does in fact act as a computer how close, performance wise, are we to being able to replicate consciousness. At a first glance it would seem that we are not far off. The largest ever neural network, the GPT-3, contains 175 billion parameters, comparatively the human brain contains 100-1000 trillion synapses. According to Moore’s law the processing power of a computer should double every two years, therefore it should theoretically take 22 years to catch up with this 103 difference in power. However, a recent study by Albert Gidon, et al, discovered that a single human dendrite (branches on nerve cells) contains the ability to perform complex logical functions, functions which were previously thought impossible without the use of complex multi-layered neuronal structures. A feature which has not been observed in other studied animals, and possibly an explanation for our comparatively higher intelligence. This drastically increases the computational power of the human brain, exponentially increasing the processing power required to replicate it.
With some people believing that a computer program could never perceive consciousness and with Moore’s law looking like it is starting to plateau, it is looking like strong artificial intelligence is not the impending doom that Elon would have us believe. Artificial consciousness may never come to fruition or at the very least will take a great deal more time and research.
Would you like to exchange ideas with us on the subject of digital transformation and process automation without obligation? Let’s talk!