|Technology:||Big Data, Data Mining, Data Quality - Quality of data, Robotics|
Artificial intelligence (AI, engl. Artificial intelligence, AI) — science and technology of creation of intelligent machines, especially intellectual computer programs. AI is connected with a similar problem of use of computers for understanding of human intelligence, but is not necessarily limited to biologically plausible methods.
What is artificial intelligence
- (J. McCarthy) of AI develops machines in which the reasonable behavior is inherent
- (Britannica) of AI - capability of digital computers to solve problems which usually are associated with highly intellectual opportunities of the person
- (Faygenbaum) of AI - develops the intelligent computer systems having potential which we traditionally connect with human mind: understanding of language, training, capability to argue, solve problems, etc.
- (Elaine Rich) of AI is science how to teach to do computers something what at the moment the person is more successful in
The intelligence (Latin of intellectus — feeling, perception, understanding, understanding, a concept, mind), or mind — the quality of mentality consisting of capability to adapt to new situations, learning capability and storing on the basis of experience, to understanding and application of abstract concepts and use of the knowledge for management of the environment. The intelligence is the general capability to knowledge and the solution of difficulties which integrates all informative capabilities of the person: feeling, perception, memory, representation, thinking, imagination.
In the early eighties Barre and Faygenbaum, erudite in the field of the theory of calculations, offered the following determination of the artificial intelligence (AI):
|The artificial intelligence is the field of information science which is engaged in development of intelligent computer systems, i.e. systems having potential which we traditionally connect with human mind — understanding of language, training, capability to argue, solve problems, etc.|
Began to refer a number of algorithms and program systems which distinctive property is that they can solve some problems as it would be done by the person reflecting over their solution to AI later.
The AI main properties are an understanding of language, training and capability to think and that is important, to work.
AI – a complex of the related technologies and processes developing qualitatively and promptly, for example:
- text processing in a natural language
- machine learning
- expert systems
- virtual agents (chat-bots and virtual assistants)
- systems of recommendations.
AI methods: NLP, CV, Data Science
Natural language (NLP) Speech technologies
- texts: will recognize, automatically translate
- speech: will recognize, generate
Computer vision (CV)
- find, trace, classify, identify objects
- take data from images
- analyze the acquired information
It is applied for
- object recognitions
- video by analytics
- descriptions of contents of images and video
- recognitions of gestures and hand-written input
- intellectual image processing
- take knowledge
- find patterns in data
- Machine learning, Deep learning
National strategy of development for artificial intelligence
Researches in the field of AI
- Overview: Artificial intelligence of 2018
- Main article: Researches in the field of artificial intelligence
Standardization in the field of AI
Standards in the field of artificial intelligence in health care
3 main trends of artificial intelligence in 4 minutes
The fourth release of a podcast of TAdviser was written by Svetlana Vronskaya, the author of Telegram-канала Analytics Now. Its subject – the main events in the market of the Artificial Intelligence (AI) which took place in 2019.
Rosstandart approved the first standards in the field of AI
The Federal Agency for Technical Regulation of Metrology (Rosstandart) approved in December, 2019 the first national standards in the field of artificial intelligence – GOST P 58776-2019 "Monitors of behavior and forecasting of intentions of people. Terms and determinations" and GOST P 58777-2019 "Air transport. Airports. Technical means of examination. A technique of measure definition of quality of recognition of illegal investments according to shadow x-ray images".
The standard is designed to provide effective communication of the intelligent robotic systems (including pilotless vehicles) with the person. Interaction of intelligent systems consists in forecasting of intentions of each other and determination of further actions based on this forecast. The forecast of behavior can be also used for identification of people with criminal intents.
The second accepted standard, GOST P 58777-2019, sets uniform requirements to systems and algorithms of recognition of illegal contents of baggage and hand luggage according to x-ray images. The standard will also increase reliability of results of tests of systems and algorithms.
Standards are developed by Peter the Great St.Petersburg Polytechnic University and come into effect since September 1, 2020.
Experts of ISO/IEC supported the offer on development of the standard in Russian
On April 16, 2019 it became known that the subcommittee of ISO/IEC on standardization in the field of artificial intelligence supported the offer of the Technical committee "Cyber-physical Systems" created based on RVC on development of the standard "Artificial intelligence. Concepts and terminology" in Russian in addition to the basic English version.
Terminology standard "Artificial intelligence. Concepts and terminology" is fundamental for all family of the international normative and technical documents in the field of artificial intelligence. Except terms and determinations, this document contains conceptual approaches and the principles of creation of systems with the AI elements, the description of interrelation of AI with other end-to-end technologies and also the basic principles and frame approaches to normative and technical regulation of artificial intelligence.
According to the results of a meeting of profile subcommittee of ISO/IEC in Dublin experts of ISO/IEC supported the proposal of delegation from Russia on synchronous development of the terminology standard in the field of AI not only in English, but also in Russian. It is expected that the document will be approved at the beginning of 2021.
|Development of products and services based on artificial intelligence requires unambiguous interpretation of the used concepts all market participants. The standard of terminology will allow to unify "language" in which developers, customers and professional community communicate, to classify such properties of products based on AI as "security", "reproducibility", "reliability" and "confidentiality". Uniform terminology will also become an important factor for development of artificial intelligence technologies within the National technology initiative – algorithms of AI use more than 80% of the companies in perimeter of NTI. Besides, the authority will allow to strengthen the solution ISO/IEC and to expand influence of the Russian experts at further development of international standards.|
Nikita Utkin, program manager of RVC, chairman of shopping Mall Kiber-fizicheskiye sistemy
During the meeting experts of ISO/IEC also supported project development of the international document Information Technology - Artificial Intelligence (AI) - Overview of Computational Approaches for AI Systems in which Russia acts as the coeditor. The document provides the overview of the current state of the systems of artificial intelligence, describing the main characteristics of systems, algorithms and approaches and also examples of customized applications in the field of AI. The working group 5 "Computing approaches and computing characteristics of the systems of Artificial intelligence" (SC 42 Working Group 5 "Computational approaches and computational characteristics of AI systems") which is specially created within subcommittee will be engaged in development of this draft document.
|Within work at the international level of delegation from Russia it was succeeded to achieve a number of sign solutions which will have long-term effect for development in the country of artificial intelligence technologies. Development of the Russian-language version of the standard, also from so early phase – a guarantee of synchronization with the international field, and development of subcommittee of ISO/IEC and initiation of the international documents with the Russian soredaktorstvo is the base for further promotion of interests of the Russian developers abroad" — commented.|
Anton Shalaev, deputy manager of Rosstandart
Artificial intelligence technologies are widely demanded in the most different industries of digital economy. Among the pacing factors constraining their full-scale practical use — backwardness of the regulatory base. At the same time worked normative and technical base provides the set quality of use of technology and the corresponding economic effect.
In the direction the artificial intelligence of shopping Mall Kiber-fizicheskiye sistemy based on RVC conducts development of a number of national standards which approval is planned for the end of 2019 – the beginning of 2020. Besides, together with market players there is a work on formation of the Plan of National Standartization (PNS) for 2020 and further. Shopping Mall Kiber-fizicheskiye sistemy is open for offers on development of documents from the interested organizations.
2018: Development of standards in the field of quantum communications, AI and the smart city
Technical committee "Cyber-physical Systems" based on RVC together with the Regional engineering center "Seyfnet" on December 6, 2018 began development of a complex of standards for the markets of the National Technology Initiative (NTI) and digital economy. By March, 2019 it is going to develop documents of technical standardization in the field of quantum communications, artificial intelligence and "the smart city", reported in RVC. Read more here.
Influence of artificial intelligence
Risk for development of a human civilization
In April, 2017 Stephen Hawking during the video conference in Beijing which took place within the Global conference of the mobile Internet said:
"Development of artificial intelligence can become as the most positive, and the most terrible factor for mankind. We should realize danger which it represents", - he emphasized.
As the scientist in the interview to the Wired edition at the end of November, 2017 told, he is afraid of the fact that AI can replace people in general.
According to Hawking, people can create too powerful artificial intelligence which will be extremely good in achievement of the purposes. And if these purposes do not match human, then people will have problems, the scientist considers. Read more here
Influence on economy and business
Influence on labor market
Bias of artificial intelligence
At the heart of all that is practice of AI (machine translation, speech recognition, processing of texts in natural languages computer vision, automation of driving of cars and many other things) deep training lies. This subset of machine learning differing in use of models of neural networks about which it is possible to tell that they imitate work of a brain therefore they with great reserve can be referred to AI. Any model of neural network studies at big data sets, thus, it finds some "skills", but how it uses them - for creators remains not clear that eventually becomes one of the major problems for many applications of deep training. The reason is that such model works with images formally, without any understanding that it does. Whether such AI system and whether it is possible to trust systems constructed on the basis of machine learning is? The value of the answer to the last question goes beyond scientific laboratories. Therefore the attention of mass media to the phenomenon which received the name AI bias considerably became aggravated. It can be translated as "bias of AI" or "bias of AI". Read more here.
Market of artificial intelligence technologies
The market of AI in Russia
World market of AI
Scopes of AI
Scopes of AI are rather wide and cover both usual to hearing of technology, and the appearing new directions far from mass application, in other words, it is all range of solutions, from vacuum cleaners to space stations. It is possible to separate all their variety by criterion of key points of development.
AI is not monolithic data domain. Moreover, the AI some technology directions appear as new subsectors of economy and the isolated entities, at the same time servicing the majority of spheres in economy.
Development of use of use of AI leads to adaptation of technologies in classical branches of the economy for all chain of value creation and will transform them, leading to an algoritmizirovaniye practically of all functionality, from logistics before company management.
Technologies and solutions of artificial intelligence: change point. Overview of TAdviser
The so-called "strong intelligence" is practically not mentioned in material - it is possible to reach it, perhaps, only in dreams of futurologists of different sense. The speech goes about reality, and it is not less interesting and fascinating. Moreover, the expert community gives definite answer: we are in a change point, and, in the most active phase of transformation of promptly computerized society. The unjustified expectations of different sense splitting society are inherent to this point of a change, on the one hand. On the other hand, in this point the extensive prospects of real development cleaned from excess hype are offered. How do they look, in terms of possible practical implementations?
About it there was a big conversation with scientists and developers of solutions of the artificial intelligence (AI) which are found on leading edge of development of AI in the key directions:
- Computer vision.
- Speech recognition.
- Understanding of sense of texts.
- Analytics and decision support.
- The deserted and digitized enterprises.
Read in more detail here.
AI in decision making: today and tomorrow
Main article: AI in decision making: today and tomorrow
Use of AI for the purpose of defense and in military science
Use in education
Use of AI in business
AI in fight against fraud
On July 11, 2019 it became known that in only two years the artificial intelligence and machine learning will be used for counteraction to fraud three times more often than for July, 2019. Such data were obtained during joint survey of SAS company and Association of the certified specialists in investigation of plunders and fraud (Association of Certified Fraud Examiners, ACFE). For July, 2019 such the antifraud tools is already used in 13% of the organizations which participated in poll and in another 25% said what one or two years are going them to implement during the next. Read more here.
AI in power industry
- At the level of design: the improved forecasting of generation and demand for energy resources, reliability assessment of the energy-generating equipment, automation of increase in generation at jump of demand.
- At the production rate: optimization of preventive maintenance of the equipment, increase in efficiency of generation, decrease in losses, prevention of thefts of energy resources.
- At the level of promotion: optimization of pricing depending on time of day and dynamic tariffing.
- At the level of providing service: the automatic choice of the most profitable supplier, detailed statistics of consumption, the automated customer service, optimization of energy consumption taking into account habits and a customer behavior.
AI in the production sphere
- At the level of design: increase in efficiency of new product development, automated scoring of suppliers and requirement analysis to spare parts and parts.
- At the production rate: improvement of process of execution of tasks, automation of assembly lines, decrease in quantity of errors, reduction of delivery dates of raw materials.
- At the level of promotion: forecasting of volumes of provision of services of support and service, management of pricing.
- At the level of providing service: improvement of planning of routes of the park of vehicles, demand for vehicle fleet resources, quality improvement of training of service engineers.
AI in banks
- Chat-bots and voice assistants
- Personalisation of products and offers
- Image identification - is used including for recognition of clients in departments and transfers of specialized offers to them
- Robots assistants
- Detection of a fraud
- AML & KYC
- Credit ratings
- Risk management
- Document handling
AI on transport
Main article: Artificial intelligence on transport
- Auto pilot (unmanned vehicle)
- The autoindustry on a revolution threshold: 5 calls of an era of pilotless driving
AI in logistics
Main article: Artificial intelligence in logistics
AI in trade
- Image identification - is used including for recognition of clients and transfer of specialized offers to them.
- Engineers of Facebook laboratory on a research of artificial intelligence provided results of work on the bot able to lie and bargain with people in June, 2017. According to Quartz, during training a system used more than 5.8 thousand real human dialogs during the negotiations collected by means of the crowdsourcing Amazon Mechanical Turk online platform.
Analyzing dialogs, a bot not only learned to conduct negotiations — he learned to lie. Engineers note that the artificial intelligence can imitate interest not to what really interests him, and to other subject. Having won preferences during negotiations, it returns to an interest object. However, engineers are up to the end not sure whether the bot learned lies from human dialogs, or came to tactics accidentally, in the course of self-training.
AI in the market of objects and services of luxury
In November, 2017 the Financial Times (FT) edition published article about how the artificial intelligence (AI) in a root changes the market of luxury. Having inspired by progress of Amazon, Google and other technology giants, hour and jewelry brands address artificial intelligence to win clients. For example, virtual interlocutors on the basis of messengers can help brands to collect data of users, without violating at the same time the European law.
In March, 2017 at Baselworld exhibition of clocks the hour and jewelry de Grisogono brand provided the chat-bot which helps buyers to select ornament from gemstones from the Crazymals collection. The interlocutor tells about himself, asks clients about tastes, and then offers jewelry at choice.
In the winter of 2017 of de Grisogono provided the first IT product — the concierge service Botler which represents something between the chat-bot and "butler" who performs function of the guide on the Swiss ski resort Saint-Moritz.
According to the cofounder of Southpigalle Olivier de Coente, the difficult economic situation and the growing competition forced luxury brands to implement innovations, including the systems of virtual interlocutors to hold clients.
According to results of the research conducted by Facebook in 2016 the bigger 50% to respondents are more convenient to send text messages, than to call in customer support. The research showed that monthly different companies received over 1 billion messages.
According to FT, AI not only improves customer interaction, but also is an important source to information for brands. As information is provided by the client directly, these data are more useful, than information obtained using cookie files or the history of viewings.
In combination with other data, such as profile of social networks of the user and demographic data, brands will be able to understand better trends, emotions and mood of clients and respectively to adjust the strategy of product management.
In the summer of 2017 the Montblanc company provided the Summit smartwatch equipped with the assistant to Google on the basis of AI. Despite classical appearance, model performs functions of the navigator, translator and assistant with voice actuation.
Within the SIHH exhibition in Geneva in January, 2017 the Jaeger-LeCoultre brand invited everyone to test the developments. Issued to visitors a bracelet with the QR code and asked to select hours using the application for iPhone. This technology on the basis of augmented reality helped the company to collect customer information.
AI in brewing
Use of AI in a state administration
AI in housing and public utilities
Problems of machine learning:
- forecasting of technical condition of the house (elevator, roof)
- forecasting of a consumption of water and electricity (regression), prediction filling of indicators (classification)
- recognition of a photo of counters
AI in criminalistics
- Image identification - is used including for identification of criminals in public spaces.
2018: Use of AI by the Dutch police for investigation of difficult crimes
In May, 2018 it became known of use of artificial intelligence by the Dutch police for investigation of difficult crimes.
According to the The Next Web edition, law enforcement agencies began to digitize more than 1500 reports and 30 million pages connected with unsolved affairs. Transfer materials to a computer format, since 1988 in which the crime was not solved not less than three years, and the criminal were it is sentenced to more than 12 years of imprisonment.
After digitization of all content it will be connected to the system of machine learning which will analyze records and to decide in what affairs the most reliable evidence is used. It should lower time of processing of affairs and disclosure of last and future crimes from several weeks to one day.
The artificial intelligence will distribute affairs on their "resolvability" and to point out possible results of examination of DNA. Then it is going to automate the analysis and in other areas of judicial examination and, perhaps, even to cover data in such areas as social sciences and testimony.
Besides, as Jeroen Hammer told one system developers, API functions for partners can be in the future released.
|Now very few people are engaged in it, and, as far as I know, nobody works with unsolved affairs. Actually, there are police officers at whom in an outlook calendar the annual reminder that it is necessary to call National judicial Institute is configured and to learn whether some new methods of the analysis of physical evidences appeared.|
In the Dutch police there is a special division specializing in mastering of new technologies for disclosure of crimes. It also created an AI system for quick search of criminals on proofs.
AI in the judicial system
Developments in the field of artificial intelligence will help to change cardinally the judicial system, to make it more fair and free from corruption schemes. Such opinion was expressed in the summer of 2017 by the Doctor of Engineering, the technical adviser of Artezio Vladimir Krylov.
The scientist considers that the solutions which are already existing now in the field of AI can be applied successfully in different spheres of economy and public life. The expert specifies that AI is successfully applied in medicine, however is in the future capable to completely change also the judicial system.
"Daily browsing news messages about developments in the field of AI you are only surprised inexhaustibilities of imagination and fruitfulness of researchers and developers in this area. Messages about scientific research constantly alternate with publications about the new products rushing on the market and messages about the surprising results received using use of AI in different areas. If to speak about the expected events accompanied with noticeable hype in media in which AI will become the hero of news again then I, probably, will not risk to do technological forecasts. I can assume that emergence somewhere of extremely competent court in the form of artificial intelligence, fair and incorruptible will become the next event. There will be it, probably, in 2020-2025. And processes which will take place in this court will lead to unexpected reflections and aspiration of many people to transfer AI the majority of management of human society".
The scientist recognizes use of artificial intelligence in the judicial system "a logical step" on development of legislative equality and justice. The machine mind is not subject to corruption and emotions, can accurately adhere to a legislative framework and make decisions taking into account many factors, including data which characterize disputing parties. By analogy with the medical sphere, robots judges can operate with Big Data from storages of public services. It is possible to assume that the machine intelligence will be able quickly to process data and to consider much more factors, than the judge-person.
Experts-psychologists, however, consider that the lack of an emotional component by consideration of legal cases will negatively affect quality of the solution. The verdict of machine court can be too rectilinear, not considering importance of feelings and mood of people.
AI in sport
AI in medicine (health care)
Analysis of behavior of citizens
Use of AI in cultural development
Work of media and literature
How do robots replace journalists, writers and poets?
Main article: Artificial intelligence and creation of music
In 2015 the command of Google tested neural networks regarding an opportunity to independently create images. Then the artificial intelligence was trained at an example of a large number of different pictures. However, when the machine "asked" to represent independently something, it turned out that it interprets the world surrounding us a little strange. For example, on a task to draw dumbbells, developers received the image in which metal was connected by human hands. Possibly, there was it because at a grade level the analyzed pictures with dumbbells contained hands, and a neural network incorrectly it interpreted.
On February 26, 2016 in San Francisco at a special auction representatives of Google helped out from the psychodelic pictures painted by artificial intelligence, about $98 thousand. These means were offered on charity. One of the most successful pictures of the machine is given below.
The picture painted by artificial intelligence of Google. Source: factroom.ru
At the beginning of 2016 using similar technology the picture "Following Rembrandt" was painted. Researchers of the Next Rembrandt project during the work analyzed about 350 pictures of the great artist, using 3D - scanners which allowed neural network to catch even the smallest parts of works and to copy style of writing of all works.
At the second stage scientists defined that specifically the neural network should write independently. For this purpose selected the man of 30-40 years having vegetation on the person, a collar and a hat.
Portrait Following Rembrandt
The received results leave double feeling. On the one hand, all of us see that before us operation of the machine. With another – if we deliver a row several cloths of Rembrandt, then very few people will be able to distinguish this portrait.
2019: Microsoft provided the generator of images on the basis of works of art
At the beginning of March, 2019 Microsoft announced the project on generation of images on the basis of works of art. For creation of this service developers used microservice architecture of deep neuronets, services Azure and storage of BLOB objects. Visual Studio Code and Azure Kubernetes Service allow to create new images in real time and are responsible for interactive appearance of the website. Read more here.
2018: The portrait written by artificial intelligence is sold for $433 thousand.
In October, 2018 at Christie's auction in New York sold the picture painted by artificial intelligence. From the Bellamy's Family series which is created on the basis of cloths of the famous painters paid $432.5 thousand at the original price of lot of $7-10 thousand for "Edmond Bellamy's Portrait" Read more here.
2020: Sberbank attracts artificial intelligence to interpretation of manuscripts of Peter I
Games (go, poker, chess)
- In the summer of 2017 it became known that Microsoft Research and Maluuba, the startup in the field of deep learning purchased by corporation at the beginning of 2017 taught to play artificial intelligence in one of the most popular computer games of all times of MS. Pac-Man. And not just taught, and made of it the champion who broke the world record set by the person.
Playing the version of the well-known arcade of MS. Pac-Man which is released for one of the first home Atari 2600 consoles the artificial intelligence could gain the maximum quantity of possible points – achievement which was impossible before. The result of the smart machine was 999,990 points whereas the best result delivered by the person is equal to 266,360 points.
When training artificial intelligence the method under the name "hybrid architecture of awards" was used. It is that to 150 special programs agents the specific objective is appointed: avoid ghosts, it is correct to move, collect granules and so on. Using programs agents the artificial intelligence independently distributed priorities for achievement of the maximum result. Version of the game MS. Pac-Man for Atari 2600 was used with an ulterior motive. The code of a game in it is less predictable, than in the original version. The strategy of development was use of perspective approach of training with a reinforcement (reinforcement learning) which assumes that the algorithm is given for processing examples of desirable behavior, and it is improved by a trial and error method. According to the scientists working on the project, such achievement will make a contribution to natural languag processing and also will be able potentially to form the basis of the systems of detailed prediction of the consumer behavior caused by a set of factors.
- In 2016 the computer for the first time beat the person in goIn May, 2017 the strongest player in guo Ke Jie from China lost the second game to the AlphaGo program. Thus, AlphaGo ensured a victory in a tournament from three batches. Ke Jie, the experts monitoring a match noted, "ideally" began a batch, creating combinations, difficult for the rival, on all game field. However AlphaGo was succeeded to simplify a game and to achieve a victory.
- In the 2017th under the threat there was a poker — specialists from Carnegie's University — Mellon created a bot which threw down a challenge to professional players. The Libratus program developed at Carnegie's University — Mellon, won a 20-day poker tournament "against Brains Vs. Artificial Intelligence: Upping the Ante". The computer won counters for the amount more than 1.7 million dollars, reports].
120 thousand distributions in unlimited Texas holdy in private were played in a tournament which took place in the Pittsburgh casino Rivers (Hedz-Ap), against Libratus Danijel Macaulay, Jimmy Choo, Dong Kim and Jason played the Wood. As a result of a 20-day tournament the program defeated people, having earned more than 1.7 million dollars in counters. Despite it, the developer will not receive any money, and prize fund in 200 thousand dollars will be divided between four living players depending on the taken place.
Definitely it is not known, Libratus how exactly works, authors described only the general structure of the program and are going to publish article in the reviewed magazine in the near future. According to developers, Libratus consists of three parts. Main "core" was prepared by Libratus in advance, calculations occupied 15 million a core hours while on Claudico two-three million left. The second part of the program monitored possible errors which rivals could allow, and considered this information in the course of the game. The third part of Libratus kept track of own weaknesses which opponents could use, and adjusted general strategy taking into account these data. Such approach allowed the program both to bluff independently, and to distinguish misinformation from rivals.
According to authors of the program at systems similar to Libratus, the big future in the most different spheres where to have to deal with incomplete information. As possible scopes of the program researchers call information security, military science, auctions, negotiations and even thrifty distribution of medicines.
Poker — a game in which it is very difficult to train to play the computer: the good player distinguishes quickly the strategy put in artificial intelligence and finds a method to win a bot. Especially difficult it is necessary the computer if rates at a poker table are not limited, i.e. the player can put an unlimited number of counters in the course.
Nevertheless, poker bots — very popular direction of development of a game. There are two types of poker bots. Some are quite simple and battle against people in a game with small rates — in it the level of poker is very low, and people cannot solve even the simplest strategy. Such bots are not really interesting to science and serve for earning money — the poker websites, as a rule, try to fight with them.
The second type — bots which compete with professionals. They are necessary not only and not just for earning money how many for promotion of science. A subject of "games with incomplete information" now one of the most popular in economic science — not casually Lloyd Shapley and Alvin Roth got in 2012 the Nobel Prize on economy for the theory of stable distribution which is connected just with "games theory". If the computer steadily learns to play better than the person games with incomplete information, perhaps, we should not bargain and suffer a question of any more whether we lost, buying the new car with characteristics necessary to us at this price — because will solve for us this application in the Computer smartphone.
- Developers of computer games apply AI in a varying degree of development. It forms the concept "Game Artificial Intelligence". Standard problems of AI in games are finding of a way in two-dimensional or three-dimensional space, simulation of behavior of fighting unit, calculation of right economic strategy and so on.
On August 4, 2017 it became known that engineers of Google corporation during joint survey with scientists from the Massachusetts Institute of Technology (MIT) created the algorithm retouching photos in real time.
Developers even more often resort to the so-called "computing" photo: to the different algorithms and AI technologies designed to improve the pictures made using smartphones. According to representatives of Google and MIT, their technology will not just allow to process qualitatively photos, but also to receive the result comparable to as though the professional photographer was engaged in it.
During testing of neural network five photos created by Adobe and MIT were selected. Each image was retouched by five different photographers. Then the received pictures were used to define, how exactly it is possible to improve each image due to brightness control, saturation and other parameters.
|Images which remove modern cameras are often perceived as crude material for the photo. Before loading the photo into social networks, even those who shoot with phone are spent minute or two, aligning color and contrast with the help of popular applications, said in the blog MIT.|
"the convolution neuronet" is the cornerstone of an algorithm of Google-MIT. Computing power of modern mobile devices is insufficiently high for a full-fledged work of a system, but researchers managed to bypass this restriction — a system executes the most part of calculations on the reduced copy of the source image, and then transfers results to the photo in high resolution.
Researchers tested the program on the normal smartphone (the model is not specified), and the algorithm could issue in real time on the screen the processed image with the resolution of 1920×1080 and update rate of 40-50 Hz. The size of all software does not exceed the size of one digital photo and can be used for processing of pictures in different styles. According to researchers, the neural network can be "trained" on a new set of images to imitate style of certain photographers.
Processors for artificial intelligence
Intel is going to press Nvidia in the market of chips for neural networks
As of 2018 practically all applications anyway connected with neural networks work on servers of Nvidia company and if others, then all the same on GPU Nvidia. But there is a serious chance that the monopoly of Nvidia will be broken by efforts of Intel. The competitor capable to press, and can be even to displace GPU from the leader's position, there will be new, not having analogs Intel processors Nervana Neural Network Processor (NNP). In them, as it appears from the name, the intellectual property purchased by Intel together with Nervana company in 2016  (more detailed) is implemented.
Amazon develops AI chips
In February, 2018 it became known of development of own chips by Amazon company. They are focused on the computing tasks connected with the artificial intelligence (AI). Read more here.
- Robots (robotics)
- Robotics (world market)
- In the industry, medicine, fighting
- Service robots
- Collaborative robot, cobot (Collaborative robot, kobot)
- IoT - IIoT
- Artificial intelligence (AI, Artificial intelligence, AI)
- Artificial intelligence (market of Russia)
- In banks, medicine, radiology
- National Association of Participants of the Market of Robotics (NAPMR)
- Russian association of artificial intelligence
- National center of development of technologies and basic elements of robotics
- The international Center for robotics (IRC) based on NITU MISIS
- ↑ Stephen Hawking called artificial intelligence "the possible murderer of a human civilization"
- ↑ Artificial intelligence of Facebook learned to lie and bargain
- ↑ Artificial intelligence starts to revolutionize luxury industries
- ↑ How the Dutch police are using AI to unravel cold cases
- ↑ GoPro the Artificial intelligence for the first time overcame the professional player in go: Wired the Go game was thought up more than two and a half thousand years ago and still it is one of the most popular games in the world — on it the championships are regularly held. At first sight it is very simple: there is a field from cages and stones — black and white. Players should take the stones as it is possible the big area on a board. Nevertheless, this game was not subject many years to the computer. Until recently artificial intelligence could not beat players of the high level — masters.
- ↑ [https://www.newscientist.com/article/2119815-ai-just-won-a-poker-tournament-against-professional-players/ New Scientist AI just won a poker tournament against professional players
- ↑ the Artificial intelligence defeated professional poker players
- ↑ against the person. This time in poker the Artificial intelligence tries to beat four professional players
- ↑ The AI algorithm can replace professional photographers