Prof. Oliver Stock, an expert in artificial intelligence, is trying to create a computer that will be able to "sense" what interests tourists in the museum and guide them accordingly. The road to the human computer is still a long way from a "space odyssey", but unlike most experts, he believes that in the future there may be computers that will be smarter than us and perform actions
Yuval Dror
Artificial intelligence is one of the fields of research that captures the imagination. Since the mid-20th century, the thought that one day computers made of silicon could think like humans has attracted many researchers to the field. However, unlike the science fiction films, the study of artificial intelligence, like other fields of science, progresses with frustrating slowness, thumb on the side.
According to Prof. Olivero Stock, former chairman of the European Artificial Intelligence Organization and chairman of one of the largest research institutes in the field in Italy, those who are currently involved in artificial intelligence research know that they will not be able to build the "Hal 9000", the all-too-human computer from the movie "Odyssey" in their lifetime In space." 2001 It doesn't stop him from trying to teach his computers to invent a joke, but also to fear that one day they will become self-aware and take the reins into their own hands.
Stock, an Italian Jew, came to Israel about two weeks ago as part of the "Israel Gateway 1" conference, which aims to strengthen the ties between Italy and Israel. The Italian ITC-IRST institute, of which Stock served as chairman until about a year and a half ago and where he currently works as a researcher, also participated in the conference. The institute, which specializes in artificial intelligence and employs about 200 researchers, took advantage of the opportunity to solemnly sign a joint investment agreement with the University of Haifa. The institute's annual budget is estimated at 15 million euros, and the investment agreement with the University of Haifa will amount to 1.2 million euros, within the framework of which joint research in the field of artificial intelligence will be carried out in the next three years.
When Stock was asked what the purpose of artificial intelligence was, he was speechless. "It is not an easy task", he explains. After a few seconds of thought, he says: "The goal of artificial intelligence, in my opinion, is to make computers understand us, to understand our cognitive system, so that they can help us without having to be experts in technology."
An example of such computers is the project in which Stock is involved these days. "I'm trying to create technology that will help tourists in museums," he says. "A tourist enters the museum. He has a certain culture, a certain language and he has his interests. He fills out a short questionnaire before entering the museum and receives a handheld computer and headphones. The handheld computer will speak to the visitor in his own language, take into account his interests, notice which picture or statue he lingered at and for how long, and thus it will be able to direct him to similar things." Stock believes that such a system, especially in a country like Italy, which is fueled by tourism, has great potential.
Computer limitations
The importance of the interface, the one that connects man to machine, led Stock to focus on the field of computer interfaces. "The history of computers meant that they first focused on hardware. Only now do we understand the importance of the software and the software interface", he says and adds, "for an interface to be truly smart, it needs to understand us, understand the language and its pragmatics". According to Stock, when it comes to interfaces, computers have advantages and disadvantages. "I'm Italian so I talk a lot with my hands. I move my head when I speak, and you in response decipher my body language. It's much more difficult for a computer."
And what is the perfect computer system? "Of course not Microsoft's 'Windows'", he says and laughs. "The perfect system will understand what you want from analyzing your behavior. In fact, we are trying to develop the perfect psychologist, one who will understand what is behind our body movements, behind our thoughts and will give us an answer in real time."
Stock knows that developing such a system is a double-edged sword. "There are people who don't want us to open a too smart system, they don't want computers that understand us more than we understand ourselves. We are also not interested in a system that will harass us. A good example of unbearable bullying is the 'fastener' application that appeared in the Office version. He tried to help users, gave them advice, but basically bullied them. The best computer is a computer that you won't feel is there. He helps you, but you don't know anything about him."
Since the computers need to understand us, they need to be taught how we think. Today, the largest information system in the world is the Internet and according to Stock, the network lacks tools that could help us. "The web is a vast body of information that has no real mind. The Google search engine, for example, is a good search system, but we need a system that knows how to extract text from an article. A system that will know how to take an interview from one online newspaper, a picture from the second, a sentence from the third and an article from the fourth, and it will know how to present all this information to you without unnecessary repetitions. The only way to do this is to make computers understand what they are reading," explains Stock. Are we getting close to that? "The full depth of language is almost infinite. It's very complicated. On the other hand, great hope lies in language, because those who understand the language understand the human experience."
Stock also knows that understanding the language is not enough. The world around us is a visual world and those interested in understanding how humans experience the world should see it with their own eyes, even if they are artificial. "When we started researching the field of image analysis, we believed that it was not a complex task. Today we know that this is a complicated project. The computer needs to understand where the object begins and where it ends, it needs to move from analyzing the smallest components of the image, such as the texture of the object, to a higher perception of a concept. The problem is that the visual world, like the verbal one, is deceptive. It is ambiguous: images and words can be analyzed in different ways. That is why the software should be equipped with a huge database that consists of its experience of the world."
However, acquiring experience is not a simple task for software that cannot leave the computer. How will she be able to learn that there is a situation where you can see the horizon but it is impossible to move towards it because there is a glass window in the way? How will she learn what the experience of running into transparent glass is like? "This is where the robots come in," says Stock. "They feel the world and study it. Unfortunately, it is taking longer than we would like. But it happens. In the end, we will be able to teach them to feel as well."
Not a futuristic field
Stock is not kidding. In fact, he is trying to teach the computers to joke, to add a sense of humor to them. "We all love humor. 90% of the ads in Italy are ads that have a certain form of humor, seasoned with excessive exposure of the woman's body. If we manage to make the computer laugh at us, we can make it produce personalized advertisements that will be directed to a specific person and take into account his sense of humor."
According to Stock, research in the field is only in its early stages. "What's really funny is that the European Union agreed to finance part of the project," he says and laughs.
Stock rejects outright the claim that the field of artificial intelligence is stuck and that the aura that surrounded it has disappeared. "There are those who say that artificial intelligence does not exist, that it is a fake, because people do not see robots from 'Star Wars' or the 'God' 9000 around them. The problem is that people think that artificial intelligence is a futuristic field, which deals with what computers will be able to do. And so, even though the computers around us have become more and more intelligent, and in some cases this is due to developments in the field of artificial intelligence, people do not attribute their wisdom to progress in the field."
Another reason why the field is not progressing at the desired rate is the lack of cooperation. "In order to create an artificial creature that will be truly intelligent, a huge effort is needed that requires the investment of more resources than those required to decode the human genome. When you come to a conference in the field of artificial intelligence, you see different and diverse topics. There is no uniformity of purpose."
Towards the end of the conversation, Stoke surprises. Experts in the field of artificial intelligence are often asked if they are not afraid that one day computers will become smarter than humans. Almost everyone says that this is unlikely, since in the end those who design the computers are the humans themselves. Stock thinks otherwise. "There is definitely a danger that they will be smarter than us, stronger than us, why not? All over the world we see how people do terrible things. If artificial intelligence develops independent intentions and an agenda of its own, why wouldn't it do terrible things?" Even so, Stock remains optimistic. "The good news is that artificial intelligence of this kind will not develop in our generation. We have enough time to think about the problems that may develop and the ways to prevent them. Maybe on that occasion we will find ways to prevent people from harming other people."
