}

The Replicant's Jazz Soul

2004/02/01 Lopez de Ipina, Karmele | Ezeiza, Aitzol | Lopez de Ipina, Montxo Iturria: Elhuyar aldizkaria

They knocked on the door. Butane distributor. On Saturday we chatted, it is very nice. But what are you doing here these hours? The truth is that I don't need bombones.
- Good night! I feel coming at these hours, but... – He has put those eyes of lamb -...on Saturday I forgot the wrench here. In the conviction, we started kissing passionately. I am fascinated by what is happening. He is an ideal man! Hmmm, so sweet...
"And this pride that you have here?" I asked him caressing a scroll that has on his chest.
"It's the on/off button," he replied with a funny smile.
Images: Montxo López de Ipiña

2001: Since the HAL smart machine surprised people in the famous movie Odyssey in Space, there is a great interest in the reasoning and communication ability of computers. At first, the researchers realized the great challenge of achieving the dialogues and reasoning that appear in the film, but like most scientific achievements, they did not cede immediately and began to take steps in development in different fields with the aim of obtaining solid bases and finding new objectives and opportunities along the way. Artificial intelligence is the art of developing computers similar to those shown in films.

In general, both among scientists and among others, the debate on the ‘humanity’ of machines is widespread. Most people believe that machines will not become human beings, because it is not even necessary, because that need does not exist, but before them there are committed dreamers working for it. Despite swimming against current, the speed and success of computer evolution are favorable. A few years ago it was unthinkable that at present they had the capacity to perform certain tasks that the computers perform. In this article we will analyze the paths that are being followed to develop the intelligence and feeling of computers and open a window for the future. From this window we can see the friendly robots that will help us, the glasses that help to see the blind and the silicon students who do wrong mathematical exercises.

Since the beginning of the history of computer development, the improvement of the machines has been sought to provide them with greater capacity to perform new tasks. The English mathematician Alan Turing was the first to propose multifunctional machines. Instead of developing a machine for each function, he proposed a machine with many functions. This idea was developed as a methodology by Herbert Simon, 1978 Nobel Prize winner. Currently this methodology is called conventional method.

This method has developed machines that use rules for decision-making. The more developed the rules, the more skilled the machines will be. This approach is used, among others, in expert systems and games. These machines have specialized intelligence and try to simulate the decisions of an expert. For example, using temperature and atmospheric pressure information, you can run a machine that announces time using rules (an example rule: “If the temperature is less than 5 °C and it is cloudy will snow”). Surely this machine would be as good as experts, little.

The conventional method is complemented by the connective method. This method proposes to use as a model the functioning of neural network structures of the human brain. Machines developed with this method use the knowledge gained from experience in a given context to make decisions. In this methodology no closed rules are used, the structure is open and the system is able to learn with experience. These machines learn from the stimuli collected and can use this learning capacity in the future, with a behavior similar to the cognitive of humans.

Computers do not understand human behavior, which generates a lot of frustration for users.

In this learning process, machines, like humans, must collect the data around them, but, like humans, if they want to be able to process them, they should be able to develop emotions. Herbert Simon himself already in 1967 stressed that the theory of thought and decision making should take into account emotions. Neurologist Antonio Damasio showed that patients who had lost their physiological ability to feel emotions made painful rational decisions. The coldest aggressive executive also uses emotions in his decisions.

Most systems developed so far by artificial intelligence and computer science are based on logical reasoning. This coincides with the topic that machines are naive but very fast. With certain rules a logical conclusion is reached. Instead, human beings take into account emotions in all our decisions and in communication we have the ability to expel this element of emotion.

The structure of the human brain explains how this capacity of emotion develops in computers. According to neuropsychologists, the main structure of the nervous system, the brain, is divided into three levels: the closest part to the spine is controlled by the internal system; the upper one, which channels the hippocampus, the control of thought, perception and emotions; and, finally, the gray matter surrounding the cortex, most of the brain, performs all other functions. This section is divided into two symmetrical parts, the right and left hemispheres.

In the right hemisphere there are sections dealing with certain tasks related to the emotions of the brain: intuition, poetry, singing, rhythm, representation, creativity of new ideas, etc. It is considered a creative and sensitive hemisphere. In the left hemisphere there are parts of the tasks related to reasoning: controlled reactions, hierarchical structures, cause-effect understanding, organization, calculation, mathematical analysis, etc. It is said to be the most logical hemisphere.

One of the functions of intelligence is to warn of the danger of relying only on reasoning. Axel Munthe.

Although neuropsychologists still have controversies, it is proven that the two hemispheres of the human brain have had sections for different functions. But at the same time, deep studies clearly indicate that the interaction between the two hemispheres is high and that both participate jointly in thought processes. In general, it is known that the hippocampus, the emotional part of the brain, coordinates all stimuli that reach the brain and performs responses.

Thoughts always have a high emotional content. The androcentric theories of yesteryear argued that the hemispheres were divided as masculine and feminine, saying that women were emotional and irrational and men were more rational and clear. But in that, as in other occasions, they were wrong. Every person reasons emotionally, with the particularities of his individual character.

So far, artificial intelligence has worked more around the functions of the left brain hemisphere. Rational rules of organization and reasoning are applied and expert information is processed for logical results. Therefore, some aspects have remained unanalysed and the machines, when making decisions, do not have the wealth offered by emotions.

This approach generates great deficiencies in today's computers. Computers cannot see beyond cold data, nor use the vital experiences that emotions offer to understand human attitude. As if it were not enough, they do not understand human behavior, which generates a great frustration in computer users. Of course, whenever all computers have one of the human characteristics: that tendency to be mistaken when you least expect it, which we have often suffered all.

Fortunately, over the last few decades there has been intense work on emotional intelligence. Emotions are becoming increasingly important, which opens the way for human-like computers. Characteristics that constitute humanity or soul, capacity for natural communication, development of feelings, spontaneous reactions, unexpected behavior, etc. In addition, fuzzy logic is increasingly applied, in which values are not true or false; as in human logic, there are many mean values to choose between the two extremes. The road to the development of the natural characteristics of the machines is open and if these works are advanced robots like the androids that appear on occasion in the films will be manufactured.

Talking to Android

The use of automatic systems is increasingly common in our environment, not only in business or university environments where technological progress is everyday, but also in the home itself. The main objective of an automatic system is to facilitate the user certain tasks and improve their quality of life. The naturalness of the interaction between machines and humans is a key factor. Therefore, when the user requests an excessive effort, he yields, at least if he is not obliged to adapt to the system.

To attract users, products can offer natural oral communication. Many basic steps have been advanced and, depending on the difficulty of the work that the computer has to perform, we already have the simplest objectives available, although other tasks are still being worked. Speech treatment technology was born with the aim of providing machines with oral communication capacity, and in recent years has had a great development.

- A-BI-UD of EP-CIG. I-su-na of be-rre-hun eu-ro. The automatic ertzaina has fined me. How do I explain that I am in a hurry, that I am late home? They do not respond. If I had given you some more intelligence…

These systems have deployed their objectives in search of natural communication, successfully addressing areas of dialogue and spontaneous conversation, with minimal limits on the language used. Dialogue systems try to emulate human communication through messages with a weak reduction by running an interactive bidirectional human machine. The speaker will not meet the restrictions required by limited vocabulary systems and may use common language.

This does not happen today, for example, with telephone access to information that want to replace telephone operators. In the future, by calling the Berria Subscriber Kiosk, the operator will be able to understand phrases like “I would like to get tickets for Aimar’s party”. This type of system is called Spontaneous Speech Recognition (SSR) systems and does not force the speaker to reduce their language.

In another way, the simplest case of dialogue systems is that of question answering systems. There are advanced systems in English, but the first prototypes are still being worked on in Basque. Today we are far from being a conversation with a computer, but that research is directed and already have begun to work the characteristics, both cognitive and emotional, essential to achieve naturalness.

However, a question-response system is far from that, as Picasso said, because communication is only giving answers. The day has not yet come when machines will speak with us from the heart. In addition, sometimes human beings are able to understand themselves without words, and surely there will be difficulties to work this type of communication.


Ábacos and virtual reality

Virtual reality has a great future in multiple applications. In addition to offering the possibility of making real games, it will offer us the possibility to channel our dreams (virtual trips) and, why not, it will offer us the possibility of performing certain professional works such as a virtual surgical operation or making economic forecasts.

Virtual reality has two basic elements: scenarios and characters. In both cases, the concepts of artificial intelligence are basic to define scenarios and stories and to define characters.

Characters are defined by abacus. The abatarras are facsimile graphics, which in most cases represent the face of a character. Hindu culture has a good definition for this word: “reincarnated god”. From the creative point of view it is very interesting to integrate both definitions. How to make a graphic facsimile both a reincarnated god? The creator must transmit his soul to the abatí. In virtual reality, creating an ideal world would be impossible without feelings, without emotions, without soul. Because after all it is wanted to be a parallel world of other human beings.

Abatarras are used in many applications, especially when affective interaction with the user is required. One of them is the automatic storytelling. This line of research for the future has its pioneers, but it is still an almost uneducated series. This topic is closely related to artificial creativity. The automatic storytelling addresses the automatic, interactive and virtual storytelling.

The computer tells the story and the user actively participates in the story through his orders. In it there are several basic elements: the story or the story, the interface (sometimes the abatarras are used) and the director of the story. To channel the story is usually a planning or a story of the story, but the user can interact throughout the story.


The soul of robots

Speaking of soul is dangerous to theologians, but when we speak of the set of characteristics of humanity we use the term soul. If we focus on this, we can say that the robot Sasex has soul, at least it is soul of jazz. He is able to perform jazz melodies using artificial intelligence. This robot, thanks to its creativity, is a virtuoso playing the sax. The robot, depending on the emotions of the moment, plays the musical piece with different feelings. For many music lovers, that's more than what Kepa Junkera does.

Other examples are the dancer robot developed by Sony or the dog robot. The final product of this house, the robot QRIO, moves on two legs and has many human characteristics, since it is designed to become a friend.

Finally, there is a new science that can make machines aware: Bioinformatics. This science aims to use programmed chips that have replaced living cells. However, it is not a question of replacing human beings with machines, but of developing increasingly natural robots.

Things have advanced a lot, technology helps us a lot now, but humanity is missing. In everyday life we are surrounded by machines, but machines do not excite us, they do not understand us. You feel a lack of that complicity. The robot you have to hit has to be the one who knows what you need when you look at it.

Dreams and curiosity must be the engine of science. The point of view not to be missed is the integration of emotions and feelings. Emotions play a fundamental role in the decision system of human beings, which should be reflected in the decision making of machines.

Collaboration between doctors, psychologists, philosophers, technicians and artists can undoubtedly increase the development of artificial intelligence and exceed the limits that currently curb such development. Thus, one day machines may be sensitive, sweet and, who knows, have soul.



BIBLIOGRAPHY

  • Alan Turing home page.
    www.turing.org.uk/turing/
  • Rosalind Pricard.
    Affective Computing, MIT Press. Jean Berko Gleason Nan Bernstein Rather Psychographic. McGraw Hill. 2000.
  • MIT Artificial Intelligence Laboratory.
    http://www.ai.mit.edu/
  • Artificial emotional creatures project.
    www.aist.go.jp/MEL/ soshiki/robot/biorobo/shibata/aec.html

Gai honi buruzko eduki gehiago

Elhuyarrek garatutako teknologia