Essay on intelligence, consciousness and AI

Intelligence and consciousness are two concepts that have been studied by philosophers and psychologists for centuries. In recent years though, scientists are also concerned of these two topics, in a merged form of intelligence and technology, in other words Artificial Intelligence. The reason for that is that understanding intelligence and consciousness could be key to create Artificial General Intelligence, which would hypothetically exceed human cognitive capabilities. Although there have been many attempts to define those terms, they have not been very satisfactory to later thinkers. As they attempted to define those terms as well, it turns out these concepts are fundamentally hard to describe.

Intelligence

The human species have been obsessed with intelligence, especially about the relative intelligence. For some reason people tend to search for reasons to find out how intelligent they are. It is supposedly because of a narcissistic character of the human species. They tend to call themselves as the most intelligent and capable animals in the world. 

One way intelligent is measured, or is being tried to measure, is by the relative performance on IQ tests, but one problem that occurs when comparing results of these tests, is that the result is a number that tells something about the relative performance on the test and it does not give an absolute idea about the intelligence. When different results from other tests, taken by the same agent, are being compared, different results show up. A measurement should in theory not vary. 

When a certain mass is being weight on the moon, it is precisely possible to measure its weight on earth, whilst these tests suggest that performance is relative, whilst the intelligence remains the same. When a certain measurement in the example gives a wrong result, it is sure that the measurement tool is wrong or not precise enough, maybe even broken. Does this mean that the agent, that takes these tests is one day 10 IQ less smart than the other day?(1)

It could be argued that the different test results are a result of certain parameters, like amount of sleep etc., but it doesn’t really matter how precise or how wrong these so-called intelligence measurement tools are, because in the example of the mass, it is possible to explain the reason for different weights, as the gravity is different on the planets, it is even precisely possible to measure these gravities, and also calculate the masses of the celestial bodies based on that information. 

It is more important to firstly understand what you are measuring, because the information that will be obtained, will have meaning. If the tests are given to a fish for example, the fish will surely get the lowest point available, whilst it supposedly has a much more complex life than the human life. (Take into consideration that the behavior and thoughts of underwater creatures are not totally understood by biologists)

Consciousness

In medicine a patient is considered unconscious when the patient is not aware of its environment and does not respond to it. Awareness is being achieved by the constant impulses of receptors that are in touch with the environment or they record changes in it. In other words, basic chemical reactions used by cells are how living creatures are aware of their environment. 

Chemistry seems to be everywhere though, does this mean that everything is conscious? When in a hot cup of water, another cold cup of water is being poured, the water reacts to it and changes it temperature accordingly. Does this mean that the water is aware of its environment the same way living creatures like humans do?

There is a key difference between the way humans react to their environment and the way matter in nature reacts to its environment. The difference being that matter obeys to the laws of nature directly, there is no decision being made by the water to change the way it moves. (0 Kelvin, or no energy is when the atoms do not move, the more energy or temperature it gets, the more the atoms move) When humans react to their environment though it is either a reflex or a decision to make a certain contraction or move.  Both acts, whether it is autonomous or not, are a decision by the central nervous system.

In other words, consciousness is the ability to observe and react to the environment by a central decision taker. When matter reacts to its environment, it simply obeys, it has no means to achieve a certain goal by not obeying it, simply because it has no brains to think or even have emotions about goals. The living creatures have central decision takers that try to stay alive and reproduce, every other action is either directly or indirectly related to this goal. 

Humans do have consciousness, as we are even capable to think of complex concepts such as consciousness by simply observing our own behavior. Without a doubt animals with central nervous systems or creatures that have similar capabilities, whether it is more complex or not, have consciousness as well as humans do. The question is however if this consciousness is the same as that of humans. Would an animal or other living creature think about complex concepts like consciousness the way humans do? Does the complexity of these conscious thoughts and acts have something to do with intelligence? 

Let’s take the IQ test into consideration in order to illustrate the relation between intelligence and consciousness. Even though IQ tests, as mentioned before, are not a precise way to measure intelligence, it surely is a way to tell if something is intelligent in a certain way. Being able to solve certain questions, is certainly conscious behavior. As the environment the agent must take the test is text and its goal is to solve questions. The input is the question itself and the output or act, are the answer. 

To be able to solve those questions though, the agent must rely on information and capabilities of the central decision taker. Language it has learned, in the case of a human agent by going to school, and the thoughts it has about the question and its solution all rely on the capabilities of the central decision taker. By the information it obtained by interacting with its environment in the past, or the information that it already has in its central decision taker system, it can solve the questions. In other words, the more it behaves and learns from its behavior and environment, the more complex the questions and problems it can solve are. And this complexity, is what we call intelligence. 

Artificial Intelligence

AI contains two words: artificial and intelligence. The meaning of AI, etymologically, is an intelligence that is created. As I suggested before, it must be a conscious system or agent that must behave in a certain intelligent way. It is important to emphasize that intelligence is relative and depends on the questions, in the example of IQ tests, or certain problems it must solve. That means that the intelligence will be different depending on the goal, or problem it must solve, and the environment.

In the thought experiment known as "The Chinese Room," Searle envisions a person who is locked away in a space and given instructions on how to manipulate Chinese characters. Despite not truly understanding the meaning of the Chinese characters, the agent can use the principles to make meaningful replies to inputs in Chinese.

According to Searle, the Chinese Room thought experiment demonstrated that a machine could not comprehend language in the same manner as a human can because it is deficient in the mental processes and states needed for comprehension.

The argument can be refuted in a simple way, using the suggestion about consciousness given before. If we would apply his implication about not understanding the input, but knowing what the output must be, to the central nervous system in the human body, he implies that the human is not able to comprehend everything it does. Simply because of the reason the nervous system doesn’t understand what is happening. 

That is however the case, but the totality of the body, the nervous system, the receptors, the information in the DNA, are all components to a system that is conscious and intelligent in its totality. This means, in his thought experiment that the person that takes decisions is just one component, but the totality of inputs, thought and decision processes and outputs is what is intelligent.  

Alan Turing, on the other hand, suggested the "Turing Test," which is a test intended to test whether a computer can demonstrate intelligent behavior that is indistinguishable from that of a human. A machine could be considered intelligent if it can trick the human judge through natural language that it is a human.

This “Turing Test” could be an effective way to judge, but the problem is that it could only give the judge an idea about how humanly intelligent it is able to communicate. As I mentioned before, intelligence could also be in a different environment. This means that it could be that an AI, made with a totally different intention, but even more intelligent than a human in that context, would get obliterated by a simple language model, even though one is more intelligent than the other one. These limitations are as limiting as the IQ test is, which means a better Turing test is needed depending on the environment and the purpose.

These adapted Turing Tests would only be testing the so-called Narrow AI or weak AI.
These Artificial intelligences are only capable to work in a certain environment. The goal of AI development however is to work towards AGI, or an AI that would be able to work in a more general environment. This would mean that all these different Turing Tests together would create the true Turing test where the general competence of AI will be tested. Once AGI is capable of convincing the human judge, it will only be a matter of time exceeding the human cognitive capabilities. “The Future” is what I call everything after the Turing Test, from that point humanity should be able to create wonders. 

In summary, I suggest that intelligence and consciousness are closely related. It should be possible to recreate intelligence, as it is only a byproduct of consciousness, whilst some may argue the opposite. Some attempts to define intelligence or tests for intelligence haven’t failed necessarily, but they weren’t detailed and generalized enough for the concept of consciousness and intelligence. In the case of Searle, I suppose he was right about the assumptions, but wrong about the conclusions.

1. A. S. Kaufman, IQ testing 101.  (Springer Publishing Company, 2009).