Consciousness is a complicated concept to define, but an international team of neuroscientists has made a solid attempt to do just that, to define what exactly consciousness is and to determine if we are close to the Holy Grail of Artificial Intelligence: an AI with its own consciousness.
The short answer to this – nothing new – approach is … no. But before we complain, the experts do explain how we could build fully conscious minds.
We have two ways of raising consciousness, in a simple and in a complicated way. The simple way is difficult, so the complex one, imagine how it must be (having to define characteristics as important as free will). Despite this, we could say that consciousness is the thoughts and feelings that we all experience individually.
Assuming that consciousness itself is based on the physical laws outlined in physics and chemistry, we should be able to find a way to chisel or compose it. This was one of the dreams of the scientist Alan Turing – who wanted to achieve a machine that mimicked the human brain – who today would be impressed by the level of artificial intelligence that Google’s AlphaGo Zero has shown, for example.
As extraordinary as these AIs are, they can solve problems at unheard-of speeds, yet they still don’t know that they can solve problems. How to make them ‘wake up’?
Three types of consciousness
Scientists who have published their study in the journal Science divided consciousness into three categories: They called the lowest category C0 , equating it with solving problems that our brains do without our conscience, such as driving home from work with little or no account. In the case of computers, the example would be the vehicle without a driver, a consciousness that, therefore, they already dominate.
The next category is C1. “It refers to the relationship between a cognitive system and a specific thought object, such as a mental representation of” the fuel tank light, “the researchers write. In C1, that thought object is selected for global processing, moving it from a close relationship to one that can be manipulated in various contexts.
The last category is C2 , which acts as a supervisor who consciously observes the tasks at hand. It encompasses what we call ‘metacognition’: a sense of knowing what we know. It is also known as ‘theory of mind’.
C1 consciousness can take place without C2, and vice versa, but according to experts, neither system has an artificial intelligence equivalent. Not yet at least.
The researchers speculate that C1 evolved as a way to break the modularity of unconscious processes.
Recent advances in microchips that can store and process information like neurons could play a key role in revolutionizing existing modular technology.
To put it to work, we would need to learn more about how our own brains create their own global workspace, an architecture that gives rise to what we consider to be our consciousness.
To develop this C2 technology, the researchers suggest several processes, such as some that apply probability to decision-making, and others that have some kind of meta-memory to draw a line between what is known and what is not. Thus we could in the future build conscious machines based on our own mental ‘hardware’.
Reference: What is consciousness, and could machines have it? Stanislas Dehaene, Hakwan Lau, Sid Kouider. Science (2017) DOI: 10.1126 / science.aan8871
