Artificial Intelligence

[gL.edu] This article gathers contributions by Kyrill Olkhovsky, developed within the context of the Conceptual clarification about "Information, Knowledge and Philosophy", under the supervisión of J.M. Díaz Nafría.

Overview

The content of this paper is to examine the topic of artificial intelligence in more detail. Thereby different aspects are considered to find out the similarity of the AI systems with us humans. For this purpose, comparisons are drawn between artificial intelligence and humans or more accurately their brains. Thus, the current state of developments can be recorded. It also allows us to anticipate future possibilities.

Introduction

Artificial intelligence (AI) is currently arousing great interest in various areas of life. Its popularity is growing steadily in science, business and the media. However, AI systems bring with them many challenges that need to be considered from both a legal and ethical perspective. As AI is used in more and more sensitive areas, such as medicine, autonomous driving, or public safety, various questions must be able to be answered.

"What does it mean for an AI system to make a decision? What are the moral, societal and legal consequences of their actions and decisions? Can an AI system be held accountable for its actions? How can these systems be controlled once their learning capabilities bring them into states that are possibly only remotely linked to their initial, designed, setup? Should such autonomous innovation in commercial systems even be allowed, and how should use and development be regulated?" [1]

The answers to these questions are highly relevant for the further integration and existence of AI systems in society. Due to the fact that the executions of AI systems are currently not comprehensible and verifiable by humans, humanity still has a long way to go in terms of legal sciences and ethics. Research in this direction is long overdue. In order to determine whether AI systems can act properly from an ethical and legal perspective, the meaning of intelligence or artificial intelligence must first be clarified. In addition, the question of whether AI can ever act on the level of a human will be considered in more detail. In addition, the creation of an artificial consciousness is questioned, since we humans are characterized by this and this is currently still our unique characteristic.

Intelligence

Artificial Intelligence is an interdisciplinary field, which is why it is difficult to derive a clear definition. Therefore, we will first take a closer look at the term "intelligence". "Intelligence is the ability to learn, understand and make judgments based on reason." [2] Human intelligence is very wide-ranging and therefore runs in diverse directions with its many manifestations:

- "Musical intelligence

- Logical-mathematical intelligence

- Spatial intelligence

- Physical-kinesthetic intelligence

- Intrapersonal and interpersonal intelligence

- Naturalistic and existential intelligence

- Creative/creative intelligence" [3].

To recreate the manifoldness of human intelligence with the help of an Artificial Intelligence will still take some time. This also raises the question of whether this is even possible. In the next section, we will take a closer look at the current understanding of Artificial Intelligence. [3]

Artificial intelligence

After the definition of intelligence, we can now take a closer look at the term artificial intelligence. A universal, standard explanation for this term does not yet exist. Only different approaches have developed. Artificial intelligence can be defined as "the art of creating machines that perform functions which, if performed by humans, require intelligence". (Kurzweil, R.,1990). [2] Accordingly, the goal of Artificial Intelligence is to replicate human intelligence using software and machines. But how can the intelligence of a machine be measured?

It was back in 1936 that the mathematician Alan Turing developed a machine with which he laid the foundation on which all further research in the field of artificial intelligence was built. Subsequently, in 1950, he developed a test with which the intelligence of a machine can be measured. In this test, a human communicates with a machine. Here the test person does not know that the communication partner is a machine. After the communication, it must be decided whether the discussion partner is a machine or a human being. The machine is considered intelligent if the test person considers the machine to be a human. [6]

A major subfield of artificial intelligence is machine learning. This term includes "methods that use learning processes to identify relationships in existing data sets in order to make predictions based on them." The subfield of deep learning is subordinated to the field of machine learning. Here, artificial neural networks are used as a basis. The idea behind neural networks is to replicate the brain of a human being. [6] The comparison of the human brain with an artificial neural network will therefore be considered in more detail later on.

As already mentioned in glosariumBitri by Hainsch David, a general distinction is made between strong and weak AI systems. [5] Weak AI is the current state of the art, and its capabilities are accordingly limited. It is used to perform predetermined tasks. Here, the AI resorts to complex algorithms, machine learning and also other techniques. [2]

The AI systems depend on the fed information from the human. Therefore, any AI is only as smart as the data provided by humans. In contrast, strong AI is a supreme machine that does not yet exist in reality. This is supposed to be able to reach or even surpass the intelligence level of humans, which is why it is also called "human level AI" [2]. Whether the strong AI can really be similar to us humans, however, still needs to be investigated in more detail.

Can a Strong AI be like us?

The dream scenario of a perfect artificial intelligence created by humans is Ava from the movie "Ex Machina". She is a being that can feel and think. She is the perfect copy of a human being with mind, consciousness, will, emotions or more precisely with spirit. These beings are called Strong AIs by scientists. So far, however, they exist only in books and movies. Today's AIs are limited in their freedom, they can devote themselves to only one task. For example, the Google software AlphaGo can defeat human grandmasters in Go, but "Mensch ärgere dich nicht!" is incomprehensible to it. Likewise, it cannot distinguish a pear from an apple. In short, these artificially intelligent systems are lost in the daily routine.

Is it imaginable to build such a machine, which can integrate itself into the everyday life and feel as comfortable as we humans do? Maybe even develop the feelings, the will and the self-confidence of a human being. The easiest approach looks like this: A spectrum of simple human behaviors and abilities is taken, such as general knowledge, language, or emotional responses, and fed into a computer. Then the training begins in which the computer is supposed to learn to behave like a human being in any life situation. Occasionally, even today's computers show human behavior, but in these cases the computers were programmed to do so. If Alexa answers impudently, then only because such a behavior was intended by the programmer and Alexa was given the necessary response. This machine is quite ingenious and could pass the Turing test. As already mentioned in the chapter on artificial intelligence, the test can prove intelligent behavior that would be comparable to human behavior. However, the Turing test has a problem. It can only test whether the machine acts like a human and not whether it is a human. The simulation of human behavior is complex and some machines have already succeeded in misleading humans. For example, in 1966, the chatbot "Eliza" imitated a psychotherapist by responding to keywords with phrases. Another chat program "Eugene Goostmann" fooled a group of British subjects in 2014. People thought they were talking to a 13-year-old boy from Ukraine, who sounded even more believable due to his limited foreign language skills and lack of education. To fool people, a machine does not have to be perfect. In general, people even tend to humanize his counterpart, any pet owner would agree with me. A strong AI is defined by the fact that it can orient itself in the world, comprehend storylines and behave flexibly in the face of change, just as a human would. Creating a human machine is the next logical step. A machine that resembles a human, yet independently follows its own principles. [8]

Another good example is the movie "I Robot". It is about a robot named Sunny that follows the following three basic laws:

"1. a robot must not harm a human being or, through inaction, allow harm to be done to a human being.

2. a robot must obey orders given to it by a human being, unless doing so would violate the First Robot Law.

3. a robot must protect its existence unless doing so would violate the First or Second Robot Law."

The robot follows the three laws, but it is still able to break them and act on its own. However, we are far away from this scenario However, it is clear that a strong AI needs sensors, sensory organs and a body to act in the world. It must feel its physical, but also mental limits. It can only develop a consciousness if it understands its place in the world. What exactly is understood by a consciousness will be explained in more detail afterwards.

Consciousness

As soon as we think about something, we do so consciously; consciousness cannot be doubted. As the French philosopher René Descartes once said "Cogito, ergo sum" ("Meditationes de prima philosophia", 1641) in English "I think, therefore I am". Descartes built his entire philosophy on this famous statement. All sense impressions could be a deception, our beliefs could be errors, our whole environment could be a huge hoax. The movie "Matrix" makes an outstanding story out of this consideration, which Descartes also made. But the consciousness remains as a fact. It is controversial but the consciousness is possibly the most important characteristic of a human being. One could even go so far as to say that consciousness is the quality that makes us. Because this lets us take in experiences and process them. Through consciousness we acquire a personality of our own. A high appreciation of the phenomena of consciousness is also affirmed with statements such as that "consciousness in a highly developed form is unique to humans and not to (other) animals or artificial intelligent systems." [4] However, it is often forgotten that "consciousness" has different meanings in everyday life as well as in science. The spectrum of the term ranges from wakefulness (vigilance, sleep, coma) to self-awareness and culminates in the area of control of one's own thoughts and actions. Given the variety of conceptualizations and states of consciousness, it is not surprising that it is one of the most controversial areas of research.

The French philosopher René Descartes first gave meaning to the term conscientia "in the sense of knowledge of mental states" in 1641 and made it the central characteristic of human beings. It is also interesting to note that the term does not even exist in some languages, such as Chinese. Possibly there was no need to give this term a name. But it could also be due to cultural aspects. Who knows what would be if we had not attached the word "consciousness" to the "Cartesian conception", maybe we too would have a different understanding of consciousness. For several centuries, the existence of human consciousness has been considered a sign that the mind must operate independently of the body or brain. It is inconceivable to try to explain consciousness by "psychological and neurobiological" processes. How is it at all possible to compare such a complex construct with individual characteristics, which are only accessible by "introspection", with empirical values. At which place of our brain is the place where all our thoughts and perceptions converge and "like on the stage of a theater - are played to an inner observer?" [4].
Although the above problems exist, the topic of "consciousness" has become an indispensable part of cognitive science. Since the mid-1990s, the research area has gained great popularity. Not only in philosophy, but also in other scientific fields such as psychology or cognitive neuroscience. In the meantime, there are numerous researchers from all over the world and from a wide variety of disciplines who have developed their own theories and methods on this topic. These often intertwine and complement each other. Unfortunately, the different perspectives are only independent points on a map and cannot (yet) provide an overall picture. [4]

Consciousness influences our actions and lets us make ethically correct decisions. Now, in order for an Artificially Intelligent System to also act correctly from a human perspective, it would have to include consciousness. Therefore, the question arises whether consciousness can also be artificially created or reproduced.

Can we create Artifical Consciousness

Assuming that the human brain is a material object and subject to scientific laws, computers should be able to mimic brain processes. This is only possible if the processes have been evaluated in detail. Accordingly, intelligence and consciousness are only a process of information processing. The information processing in the nerve cells can be understood and reproduced very well nowadays. Thus, if consciousness is understood only as the result of all our brain functions, consciousness should theoretically be artificially producible. [7]

Let us assume that consciousness really can be created artificially. Then AI systems would only have to be equipped with an artificial consciousness in order to be able to make ethically and legally correct decisions.

Comparison of the human brain with the artificial neural network

In the human brain there are countless nerve cells linked together to form a network. This network is also known as the neural network and is responsible for perception, associations, thoughts, learning ability and consciousness. The human brain has been used as a model for the development of Artificial Neural Networks in Artificial Intelligence research. Artificial Neural Networks are a major subfield of AI in which attempts are made to replicate and simulate the function and also the structure of the natural networks in the brain. The structure of an artificial neural network is divided into several layers. The first is the input layer, which can be compared to the optic nerve of a human. Subsequent layers are called hidden layers, each of which receives the output of the previous layer. Here, too, a connection with the human brain can be established. Neurons that are close to the optic nerve send signals to neurons that are further away. The output layer is the last layer and transmits the result output of the AI system. Ideally, the artificial neural network should learn with each layer. [3]

The neural network in the brain is much more complex than the artificial replica. Therefore, Artificial Neural Networks are severely limited in their capabilities, which is why AI systems currently lack consciousness. But maybe one day an artificial replica of consciousness can be integrated.

Conclusion

Due to the fact that the human brain is extremely complex, it is hardly imaginable that an artificial intelligence can even begin to imitate human intelligence. The mere fact that only humans have a consciousness can set back the rapid progress in the research field of artificial intelligence. Before technical developments are pushed further, the legal and ethical perspectives should first be brought to the foreground. This is where awareness plays a major role. Until it is possible to artificially create consciousness, humans will not trust AI systems. However, a lot has already been achieved with humans as role models through imitation. Both intelligence and neural networks are being emulated with AI. Therefore, it is consistently conceivable that Artificially Intelligent Systems could one day exhibit consciousness. One thing is for sure, there is still a lot unexplored in the field of Artificial Intelligence. Here, groundbreaking developments can certainly be expected in the future.

References

[1] Virginia Dignum. (2018). Rerieved from: https://link.springer.com/content/pdf/10.1007/s10676-018-9450-z.pdf. [13.02.2018]

[2] Johannes Scherk B.Sc. (2017). Künstliche Intelligenz-Artificial Intelligence. Pächhacker Innovation Consulting.

[3] Ralf T. Kreutzer. (2019). Künstliche Intelligenz verstehen. Springer Gabler.

[4] Jochen. Müsseler. (2017). Allgemeine Psychologie. Springer. Edition: 3

[5] Hainsch David. (2018). Odyssey of Philosophy and Information. glosariumBitri

[6] Peter Buxmann. (2019). Künstliche Intelligenz Mit Algorithmen zum wirtschaftlichen Erfolg. Springer Gabler

[7] Jochen Zielke. (2018) Rerieved from: https://www.planet-wissen.de/technik/computer_und_roboter/kuenstliche_intelligenz/pwiekuenstlicheintelligenzundbewusstsein100.html. [20.11.2018]

[8] Martin Scheufens. (2019) Rerieved from: https://www.pm-wissen.com/kann-eine-kuenstliche-intelligenz-so-sein-wie-wir/. [12.01.2019]