Artificial intelligence. Current achievements and directions of development
 Скачать 0.54 Mb.
|
|
Artificial intelligence. Current achievements and directions of development. Автор Сотниченко Марина Андреевна Научный руководитель Мануйлова Наталья Степановна Координаты для связи: sotnichenko-marina@mail.ru тел. 89054865603 2017 Содержание Введение 1.Происхождение и понимание термина «искусственный интеллект» 2.Интеллектуальные возможности современных компьютеров 3.Искусственный интеллект в общественной жизни 4.Современный искусственный интеллект 5.Применение 6.Исследовательские центры 7.6 наиболее важных достижений искусственного интеллекта в 2016 году Заключение Список использованной литературы Введение Темой научного проекта является «Искусственный интеллект: перспективы развития». Актуальность данной темы заключается в важности развития технологии искусственного интеллекта для таких прогрессивных, на сегодняшний день, отраслей науки как робототехника, кибернетика и для более быстрого, удобного доступа к мировым информационным ресурсам. А также искусственный интеллект необходим для развития социальных услуг и для ведения кибервойн. Целью данного исследования является разработка предложений более эффективного внедрения искусственного интеллекта в среду обитания человека, а также составление прогнозов о возможности реализации данного проекта. Для достижения цели научного проекта были определенны следующие задачи: 1)проанализировать динамику развития технологии искусственного интеллекта за последние пять лет. 2)разработать сценарии внедрения искусственного интеллекта в человеческое сообщество. 3)определить возможные последствия после интеграции искусственного интеллекта в человеческий социум. Объектом исследования является искусственный интеллект. Предметом исследования является рассмотрения процесса внедрения и адаптации искусственного интеллекта в человеческий социум. Гипотезой исследования является вопрос о возможности внедрения искусственного интеллекта в человеческое сообщество, его адаптации в нем и его влияние на человека, и на мир в целом. Методы исследования: рассмотрение материала, проведение опроса, анализ необходимой литературы, анкетирование, теоретический анализ. Artificial intelligence. Current achievements and directions of development. The origin and meaning of the term "artificial intelligence" . Artificial intelligence (AI) — 1) science and technology for creating intelligent machines, especially intelligent computer programs; 2) the property of intelligent systems to perform the creative functions that are traditionally considered the prerogative of man. AI is associated with a similar task of using computers to understand human intelligence, but not necessarily limited to biologically plausible methods. The definition of artificial intelligence given by John McCarthy in 1956 at a conference at Dartmouth University, not directly connected with understanding intelligence in humans. According to McCarthy, AI researchers are free to use methods that are not observed in humans, if it is necessary for the solution of specific problems. Explaining his definition, John McCarthy points out: "the Problem is that while we cannot in General determine which computational procedures we want to call intelligent. We understand some of the mechanisms of intelligence and not understand the rest. Therefore, under the intelligence within that science means only the computational part of the ability to achieve goals in the world." At the same time, there is a point of view according to which intelligence can only be a biological phenomenon. As pointed out by the Chairman of the St. Petersburg branch of Russian Association of artificial intelligence, T. A. Gavrilova, in English, the phrase artificial intelligence has the slightly fantastical anthropomorphic overtones, which it acquired in quite a bad Russian translation. The word intelligence means "the ability to reason intelligently" and not "intelligence", which is the English equivalent intellect. Participants of the Russian Association of artificial intelligence give the following definition of artificial intelligence: Scientific direction, in which are put and solved the problem of hardware or software modeling those human activities that are traditionally considered intellectual. The property of intellectual systems to perform the functions of (creative) that are traditionally considered the prerogative of man. The smart system is a technical or software system able to solve problems, traditionally considered to be creative, belonging to a particular subject area, knowledge of which is stored in the memory of the system. The structure of intellectual system consists of three main components, a knowledge base, solver and intelligent interface allowing communication with a computer without special software for data entry. A science called "Artificial intelligence" is part of the complex computer Sciences, and created based on it technology to information technology. The task of this science is the reconstruction by computer systems and other artificial devices of reasonable arguments and actions. One of the particular definitions of intelligence, common for humans and "machines", can be formulated as follows: "Intelligence — the ability of the system to create in the course of self-study programs (primarily heuristic) for solving problems of a particular class of complexity and to solve these problems". Theory is not explicitly defined what constitutes necessary and sufficient conditions for achieving intelligence. While on this subject there are several hypotheses, for example, the Turing test or the hypothesis of Newell — Simon. Usually, the implementation of intelligent systems suitable from the point of view of modeling human intelligence. Thus, in the framework of artificial intelligence there are two main directions: 1.symbolic (semiotic, downward) based on modeling high-level processes of human thinking, on the representation and use of knowledge; 2.Neurocybernetics (neural network, the upside) is based on modelling individual low-level structures of the brain (neurons) . Thus, the most important task of artificial intelligence is to build intelligent computer system that would have the level of effectiveness of the solutions of non-formalized tasks, comparable to or exceeding those of a human. The criterion and constructive definition of intelligence proposed a thought experiment known as the Turing test. In the modern formulation, we can consider this task as a task of approximation of the singularity in its swarmintelligence understanding. At the moment there are no artificial intelligence systems that uniquely meet the basic objectives outlined above. Advances in the study of analog and reversible computing will make a big step forward in the construction of artificial intelligence systems. The most commonly used in the construction of systems of artificial intelligence programming paradigms — functional programming and logic programming. From the traditional structural and object-oriented approaches to the development of programme logic they are different nonlinear-and low-level support for analysis and synthesis of data structures. The AI problem associated with resources Messages about the unique achievements of professionals in the field of artificial intelligence (AI) that promised unprecedented opportunities, disappeared from the pages of popular science books, many years ago. The euphoria associated with the first practical success in the field of AI went very quickly, because to move from experimental studies of computer models to solving the applied problems of the real world was much harder than anticipated. On the difficulties of this transition drew the attention of experts around the world, and after detailed analysis it turned out that almost all the problems associated with the lack of resources of two types: computer (computing power, capacity, operational and external memory) and human (knowledge-based intelligent software development requires the involvement of leading experts from different fields of knowledge and long term research projects). To date, resources of the first type came (or will come in the next five to ten years) to the level of allowing the AI systems to solve extremely challenging practical problems. But with the resources of the second type of situation in the world deteriorates - that is why the achievements in the field of AI are associated mainly with a small number of leading AI centers at major universities.  Modern achievements in the field of information and telecommunications technologies contribute to the active development of artificial intelligence. Of course, the creation of machines possessing a fully human consciousness, speech does not go yet, however the available results allow to say now that the era of artificial intelligence is not far off. In January of this year, the founder of the world economic forum in Davos, Klaus Schwab, called artificial intelligence one of the main drivers of the fourth industrial revolution. This fourth industrial revolution is coming at us like a tsunami, and its main driving force comes from the achievements in the field of artificial intelligence, robotics, nanotechnology, Internet of things and other fields of science Klaus Schwab Indeed, betting on artificial intelligence every year doing more and more market players – ranging from major IT giants to small startups. It is worth noting that among them are not only such well-known IT players like Google, Facebook, Microsoft, IBM and the company, which, at first glance, must be far from this theme. Even General Motors and Boeing have created a joint laboratory associated with the study of artificial intelligence. In other words, artificial intelligence is becoming a kind of mainstream of our time. The total cost of the development of highly intelligent machines are estimated to hundreds of millions of dollars a year, and recent developments show that such amount is justified. Intelligent machines is growing every year, and some of them already can in certain areas surpass human capabilities. Ellectual abilities of modern computers In 2011, the IMB's supercomputer Watson managed to win the TV show Jeopardy (the Russian equivalent – "jeopardy"), in which he not only had to find the correct answers to the questions but also to "understand" their meaning in natural language. Jobs used the metaphor, and the questions were asked without any simplifications and clarifications. To cope with these challenges, Watson has demonstrated a really unique analytical capabilities, which are now used in various fields of industry. One of the greatest achievements of Watson is the ability to independently conduct medical tests. Analyzing huge data sets with different histories, diagnoses and courses of treatment and making based on the objective findings, a supercomputer proved itself a good doctor. In particular, Watson copes with the diagnosis of cancer. Project Watson for Oncology is now used in practice in several medical centers and hospitals. Among them, international hospital Bumrungrad (Thailand), new York center for the study of the human genome, etc. If we talk about the intelligence of computing machines, the real breakthrough in this area occurred just a couple of months ago when Google AlphaGo managed to beat the game th it the absolute champion Lee Sedalia. This theme was actively discussed, because skeptics believed that this is impossible in principle (at least with the current level of technology). He is really highly intelligent game in which there can be billions of combinations of chips on the Board. The players themselves not just calculate the possible combinations, but also act intuitively. Therefore, working only on the basis of busting all kinds of moves, the computers were able to beat intermediate level players but not professionals. To achieve victory in th above her champion let the technology "deep machine learning" (deep learning), which is now the most trending direction of development of artificial intelligence. The basis for this technology back in 1959 identified Arthur Samuel, when I was offered to work on the training computers, definitely not using programmed algorithms. The computer learns from examples and your own experience. The program AlphaGo first analyzed 30 million moves of professional players, and then began to play with herself. Having played millions of parties, the program learned to evaluate the best position of the stones on the Board to achieve victory . Technology of machine learning allows to model really complex non-linear relationship consisting of many layers, which significantly increases the learnability of artificial intelligence. Therefore, the scope of this technology is not limited to the game of go. It is an integral part of research in the areas of speech recognition and imaging and unmanned vehicles, and is also used for solving other complex tasks. Facebook, for example, is actively developing the technology of deep machine learning to create social networks of the future. It is expected that the developed system will be able to understand the messages in the newsfeed and show the user only the news that really interest him. The first research results are already used in practice, the algorithms Facebook.  Champion Go game Lee SEDOL analyzes their mistakes after defeat against AlphaGo. Artificial intelligence in public life The scope of artificial intelligence is limited only to the creative powers of man. At the moment, the machine learning to read the various template documents (legal, financial, etc.), to compose music, to understand people's emotions by facial expression, writing literary works and much more. So, a member of the IDC Media Innovation Lab guy Hoffman introduced the robot, independently engaged writing different pieces of music. Only a few years ago the idea was proposed to use artificial intelligence (or rather, its inverse) in the most popular gadget of our time – the smartphone. The result is a highly intelligent voice assistants, which are now used by millions of people. It is possible to allocate such well-known applications like Siri from Apple, Echo from Amazon, Cortana from Microsoft and Google Now from Google. Highly intelligent voice assistants are designed to help the user in his daily activities. For example, Google Now analyzes the user's habits by tracking all his actions and movements, and the further in advance about upcoming events. So, before going to work, the user is provided with a report on the traffic situation in the city and offers the most optimal route. Currently, active work is ongoing with the aim to make highly intelligent voice assistants, personal secretaries, who will communicate with us in natural language and perform various voice commands (send email, schedule meetings, reserve a table at a restaurant, record on reception to the doctor, etc.). The current results show that such systems will be developed very soon. This year, the developers of artificial intelligence Viv at the conference could use voice commands to order pizza. And this was done without a single call without using the keypad. Achievements in the field of artificial intelligence is also actively used in the concept of smart home. The concept has been discussed for several decades, but its practical realization became possible only recently. At the moment there are some projects that allow to control TV, air conditioning, stereo, remote unplug any electric device, configure the lighting to the rhythm of the music, etc. Developments of this kind do quite a lot of companies, among which is IT-giants. For example, mark Zuckerberg, founder of Facebook, stated that one of its objectives is the creation of artificial intelligence that is able to effectively help him around the house. I'm going to teach my assistant to understand my voice, to control at our house all music, light, temperature and so on. Teach him to know the faces of my friends when they call at the door Mark Zuckerberg . Another area where we gradually permeates artificial intelligence, are unmanned vehicles. Many experts agree that very soon the streets of all the cities will be filled intelligent vehicle, the control process which will not require human intervention. Today the owners of premium cars can take full advantage of features such as shadowing the rows of the movement, collision alert and adaptive cruise control. The modern system of car control can throttle and brake. Develop self-driving cars Google has already traveled on U.S. roads in total about 1 million miles without a serious accident. And this is only the beginning. Of course, in the development of unmanned vehicles there are a number of issues, such as behavior of such vehicles in critical situations, are quite difficult to predict. However, researchers are actively working on it, planning to solve problems of this kind in the near future. Artificial intelligence is gradually becoming a part of our lives. Did You know that some news notes You have to read, written by robots? For example, in Associated Press (AP) robots written thousands of financial reports on the state of American corporations and make it much faster than normal journalists. The application of automated technologies has enabled the Associated Press to increase the number of quarterly news from 300 to 4400. At the end of 2015, the company Yandex has also created its own news Agency, where news are written with the help of artificial intelligence. While this is only notes on traffic and weather, however, the representatives of the company say that the list of topics will be further expanded. Quite creative to use artificial intelligence approached the Japanese company Deep Knowledge, which became the first organization in the world, including artificial intelligence to your Board of Directors. Analyzing large amounts of data, a robot named Vital capable of finding trends that people might not notice. This allowed him to get the right to vote on all financial decisions of the company. Artificial intelligence is used not only for peaceful purposes. Significant funds are invested in the creation of combat robots, unmanned vehicles and military games, and it happens in many countries around the world. Among them Russia. The government allocates huge funds in the defense industry, an important direction of which is military robotics. The main objective of this industry is to create robots that can operate in the most life-threatening human conditions. Not to mention that this kind of research is being conducted quite successfully. Recently, Vladimir Putin has demonstrated a system to control robots via brain impulses.  the system of robot control using the electric signals of the human brain has been demonstrated to Vladimir Putin To development in the field of robotics for the defense industry was on the optimal schedule, according to the presidential decree of 16 December 2015 N 623 the national centre for technology development and the basic elements of robotics. The funding of this centre by the Foundation for advanced studies. As we can see, the future of artificial intelligence is already very close. Current developments show that the developed intelligent machines is growing every year. And let them study took more than two decades, which once said Turing, man today stands on the threshold of a fundamentally new discoveries in this field, which will allow to realise his dream. Modern artificial intelligence ASIMO is an Intelligent humanoid robot of Honda company. It is possible to allocate two directions of development of AI: 1.the decision of the problems connected with the approach of specialized AI systems to human capabilities and their integration, which is realized by human nature. 2.the creation of an artificial mind that represents the integration of the created AI systems into a single system capable of solving the problems of humanity . But at the moment in the field of artificial intelligence has been observed the involvement of many subject areas that have more practical relevance to AI, but not fundamental. Many approaches have been tried, but to the emergence of artificial intelligence no research team has not yet approached. Below are just some of the most famous developments in the field of AI. Аpplication Some of the most famous AI systems: 1)Deep Blue defeated the world chess champion. Match Kasparov against the supercomputer did not satisfy neither the computer scientists nor the players, and the system was not recognized by Kasparov (see human vs computer). Then the line of IBM supercomputers emerged in the projects brute force BluGene (molecular modeling), and modeling the pyramidal cells in the center of the Swiss Blue Brain. 2)Watson is a promising development by IBM, able to understand human speech and produce probabilistic search, with the use of a large number of algorithms. To illustrate the Watson participated in the American game "Jeopardy!", analogue of the "My games" in Russia, where the system managed to win in both games. 3)MYCIN is one of the earliest expert systems that could diagnose a set of diseases, and often exactly the same as the doctor. 4)20Q — project based on the ideas of AI, based on the classic game "20 questions". Became very popular after the appearance in the Internet site of the 20q. 5)The speech recognition. Systems such as ViaVoice is able to serve consumers. 6)Robots in the annual RoboCup tournament compete in a simplified form of football. Banks use artificial intelligence systems (FIC) in the insurance activities (actuarial mathematics) with the gambling, and the management of the property. Pattern recognition methods (including more complex and specialized, and neural networks) are widely used in optical and acoustic detection (including text and speech), medical diagnostics, spam filters, anti-aircraft systems (definition of targets), as well as to ensure issues of national security. The developers of computer games use AI in varying degrees of sophistication. This forms the concept of "Game artificial intelligence". Standard tasks of the AI in games is finding paths in a two-dimensional or three-dimensional space, the simulation of the behavior of combat units, the calculation of the true economic strategy, and so on. Research centers The largest scientific and research centers in the field of artificial intelligence: -United States Of America -Massachusetts Institute of technology -Research Institute of machine intelligence -Germany -German research center for artificial intelligence -Japan -National Institute of modern industrial science and technology (AIST) -Russia -Scientific Council for methodology of artificial intelligence, Russian Academy of Sciences 6 the most important achievements of artificial intelligence in 2016 2016 was the year of artificial intelligence. Progress in the area of self-management systems, voice recognition and deep learning has enabled computers to make a series of major breakthroughs that were previously impossible. Here are six of the most important of them. 1. AlphaGo defeated the world champion in the game of Go . Th is considered the most complicated game ever invented. It offers an incredible number of possible actions, and in many ways, according to the players themselves, relies on human intuition. Artificial intelligence AlphaGo taught myself the game by playing millions of parties with their copies and in March of this year managed to defeat the champion Lee Sedalia in four games out of five. 2. Tesla self-driving car was taken to hospital of a person with a heart attack Unmanned are the subject of hot debate worldwide. However, Tesla CEO Elon Musk stresses that cars with autopilot system safer than cars without it, regardless of the point of view of the public. Report of the National security Council, the United States indicates that in 2015 the number of traffic fatalities was 1.3 cases for every 100 million miles travelled by conventional vehicles, while the indicators of Tesla – 130 million miles and only one was an accident. At the same time on account of the autopilot of the company at least one life saved, the car delivered his master Joshua Nellie to the hospital when in the road he suffered a sudden heart attack. 3. Swarm intelligence predicted the results of the Kentucky Derby In may, artificial intelligence UNU was able to successfully predict the four winners of the prestigious equestrian competition, and, strictly in the order of their finish. This failed to make any official expert on these competitions – so the computer won with a bet of 540 to 1. UNU was developed by company A. I. Unanimous, led by a specialist in human computer interaction Luis Rosenberg. 4. Microsoft AI today understands human speech better than the people themselves In October of this year, Microsoft has demonstrated that AI for the first time caught up with the man in the efficiency of automatic speech recognition. To achieve this result, the system the company used the so-called high-precision and recurrent neural networks. To prepare for the test she needed 2000 hours of recorded data. 5. AI predicted the results of presidential elections in the United States The results of the election came as a surprise to many, including insiders of the political system. However, the Indian startup MogIA in Mumbai confidently predicted victory trump. AI companies analyzed 20 million social media and was able to determine the real sympathies of voters. Although many experts are reluctant to give the victory MogIA too much importance to the objective — this intelligence was able to accurately predict an event that came as a surprise to billions of people. 6. Artificial intelligence has revolutionized the diagnosis of cancer Health care is one of the areas in which the successes of AI have the greatest practical value. In particular, the supercomputer of IBM Watson already able to see deviations in the health of the individual, which elude the attention of experienced diagnosticians. Statistically, about 30 % of cases Watson puts patients with additional diagnosis missed by medical people. Even more impressive results achieved AI the Houston Methodist research Institute in Texas. Artificial intelligence explores the millions of mammograms (the speed of analysis is 30 times more than human) and gives oncosolutions with 99% accuracy. Заключение Знаменитый нидерландский ученый, Эдсгер Вибе Дейкстра сказал: «Вопрос «Могут ли машины мыслить?» имеет примерно такой же смысл, как вопрос «Могут ли подлодки плавать?». Этой цитатой мы подводим итог нашего исследования. Изучив все достижения, человечества в создании искусственного интеллекта, мы пришли к выводу, что он уже создан. Но по каким-то причинам большая часть нашего общества не признает этого. Мы, люди думаем, что искусственный интеллект можно тогда назвать интеллектом, когда он сможет вести себя как человек. На этом основывается знаменитый тест Тьюринга. Но с чего мы взяли, что искусственный интеллект должен вести себя как человек, а не как машина. Он превосходит нас в интеллектуальном плане и физическом, он всегда объективен, он не зависит от чувств и желаний. Так почему же мы пытаемся привить нашему творению, те характеристик человека, которые делают нас уязвимыми. Эти свойства нашего вида достались нам от первобытных предков, они наши «рудиментарные органы». Так зачем, что уже не нужно нам, прививать искусственному интеллекту. Ведь мы же произошли от обезьян, согласно Дарвиновской теории эволюции, но мы же не живем в лесах, не лазаем по деревьям. Мы как вид сформировались тогда, когда поняли, свою значимость, свое превосходство над приматами, хотя произошли от них. Мы пришли к выводу, что искусственный интеллект создан на 99,9% и лишь 0,1% отделяет его от полноценного звания разумного существа. К этим вывода мы пришли в результате нашего исследования, они обосновываются на тех достижениях науки, что были описаны выше. Искусственный интеллект уже обзавелся телом, он может видеть, слышать, говорить, помнить, испытывать эмоции. Дело осталось за малым, он должен осознать свою разумность, свой интеллект, само понимание своего существование. И тогда те функции, что дал ему человек, в виде умения говорить, выражать эмоции и д.т., перестанут быть имитацией, а станут полноценными свойствами искусственного интеллекта, как новой ветви эволюции. Что же касается его адаптации в среде обитания человека, его сосуществование с ним, это вполне реально. Искусственный интеллект в виде роботов или программного обеспеченья нашего ПК, или в других формах проявления проникнет в нашу жизнь и станет ее неотъемлемой частью. Этот процесс произойдет, это лишь вопрос времени. Список использованной литературы: 1.Джордж Л.Ф. «Искусственный интеллект: стратегии и методы решения сложных проблем», 4-е издание. Пер. с англ. - М.: «Издательский дом «Вильямс», 2003 г. - 432 с. 2.Масленникова О.Е., Попова И.В. «Основы искусственного интеллекта», Магнитогорск, 2008 г. - 282 с. 3.Веб-сайт Hi - News.ru 4.Веб-сайт Hi - News.ru 5.Веб-сайт Southkoreanews.ru 6.Веб-сайт Globalscience.ru 7.Интернет портал «Инсай - Вертикальное развитие», 8.Официальный сайт газеты «Деловой Казахстан», 9.Интернет портал «Tengri News» 10.Непомнящий Н.Н. «Сто великих загадок 20 века», М.: ООО «Издательский дом «Вече», 2009 г. - 470 с. 11.Софдж Э. // «Можно ли доверять роботам?: Зависимость от машин». М.: «Популярная механика», 2010 г. С. 5-6. АННОТАЦИЯ НАУЧНОЙ РАБОТЫ 1. Название работы: Artificial intelligence.Current achievements and directions of development 2. ГРНТИ1: 12.09.11. 3.Вуз: федеральное государственное бюджетное образовательное учреждение высшего профессионального образования «Донской государственный технический университет». 4.Год завершения работы: 2017 5.Объем работы: ____16_стр. 6.Количество приложений: __0__стр. 7.Количество иллюстраций: __3___ ед. 8.Количество таблиц: ___0___ед. 9.Количество источников литературы: ___11__ед. |