[an error occurred while processing the directive]
RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2

Artificial intelligence (AI) Artificial intelligence (AI)

Product
Technology: Big Data,  Data Mining,  Data Quality,  Robotics

Content

Artificial intelligence (AI) is the science and technology of creating intelligent machines, especially intelligent computer programs. AI is associated with the similar task of using computers to understand human intelligence, but is not necessarily limited to biologically plausible methods.

What is artificial intelligence

Difficulties in defining the concept of AI

A separate TAdviser article is devoted to options for defining the term AI and related difficulties - "The term AI has been used for 70 years, but everyone understands it in different ways. What is AI really? "

AI criteria

For 2024, the following criteria and features were identified that distinguish AI from other software complexes:

  • The ability to learn and adapt - the ability to independently learn and adapt based on new data, experience and feedback, optimize your algorithms in the process, which allows AI to become more efficient and accurate over time. AI is not limited to primary learning and can continually evolve.

  • Autonomy of decision making - the ability to go beyond the boundaries of established algorithms within the focus of the problem and adaptation to new scenarios without prior human intervention. Simply put, AI is able to independently find the best solution for solving a specific problem, while ordinary programs are "locked" in the boundaries of integrated algorithms.

  • Understanding the context of complex problems - AI has the ability to understand complex, multilevel tasks and the context in which they arise, whereas in traditional programs the depth of understanding is limited exclusively by pre-written and implemented scripts and algorithms.

  • Cognitive functions - information perception, reasoning, learning and multi-vector problem solving sets AI apart from any other systems, even the most complex. Logical reasoning assumes the ability to logically analyze information and form conclusions with building cause and effect chains.

  • Natural language processing - replacing the machine language with instructions and subsequently with the human language makes the software complex close to AI to the extent that AI is able to understand human speech.

  • Predictive analysis - AI can analyze large amounts of historical data, detect patterns and trends, and use this knowledge to predict future events or outcomes, building on patterns and probabilistic estimates.

  • Multimodality - refers to the ability of AI to analyze and integrate information from different sources or types of data (modalities). For example, a multimodal AI system may simultaneously process text, images, audio, and video.

  • Multidisciplinary - in the context of AI, it implies the application of knowledge and methods from different scientific disciplines to develop, understand and improve AI systems. This approach focuses on combining diverse scientific and technical areas of expertise to create more efficient and intelligent systems.

The integration of all 8 basic features of AI is not mandatory, since, in fact, even one of the above is enough.

Types of AI

For 2021, the researchers used the following classification of AI types:

Artificial Super Intelligence (ASI) is a hypothetical AI that can not only reproduce the maximum human ability, but even surpass it. Those who believe in ASI believe that he will gain the power to penetrate a person's thoughts and feelings in order to subjugate him to his will. See Super Intelligence: Futurologist horror stories or the real future of artificial intelligence?

Remaining also hypothetical strong, or general AI (Artificial General Intelligence, AGI) in terms of reasonableness stands a step below ASI, adherents of this type of AI are limited in their beliefs by the possibility of creating machines capable of at least performing the same actions as a person.

Weak, or narrow AI (Artificial Narrow Intelligence, ANI) allows you to see weak hints of the mind in the behavior of machines (therefore it is called weak). It is designed to run only a strictly defined narrow range of applications (so it is called narrow). In the case of ANI, no human-independent autonomous behavior or independent development is possible. Systems equipped with ANIs can only exist in the form in which they were created by man and cannot even theoretically get out of his control.

AI research

The main components of AI

Artificial intelligence is the result of the synergy of many technological, scientific and industrial achievements of the previous 100 years.

There are many factors that have influenced the expansion of AI, but there are several key ones:

One of the main drivers of the rapid development of AI was computer games and gamers who moved the progress of video cards, which made it possible to exponentially increase the computing power that later began to be used for AI projects.

The more data - the more accurate the results, so AI could not appear before sufficient computing power, Big data and a high level of Internet development appeared, but all this must be correctly interpreted and processed, i.e. algorithms are needed.

AI methods

Technological directions of AI. Deloitte data

Generative artificial intelligence

Main article: Generative artificial intelligence

Data analysis

Data Science

  • extract knowledge
  • find patterns in the data
  • predicted.

The following methods are used:

AI training

Main article: Training artificial intelligence

Deep learning

Deep Learning AI. For 2023, the following applies:

  • backpropagation,
  • Generative adversarial network (GAN)
  • convolutional neural networks (CNN),
  • recurrent neural networks (RNNs),
  • direct distribution networks (FNNs),
  • deep neural networks (Deep Neural Networks) and
  • autoencoders.

Machine learning

Main article: Machine learning

Machine Learning AI - support vector method, linear regression, logistic regression, decision trees, random forest, K-nearest neighbor method (KNN).

Reinforcement Training

Reinforcement training:

  • Reinforcement Learning is actively used in Robotics AI and Decision AI
  • Q-learning, policy-based algorithms.

Natural Language Processing

Natural Language Processing (NLP) - transformer-based models are the most powerful and innovative models for 2023.

NLPs recognize and automatically translate texts, recognize and generate speech.

Computer vision

Computer vision (CV):

  • find, track, classify, identify objects
  • retrieving data from images
  • analysis of the obtained information

Applies to

  • object recognition
  • analytics video
  • descriptions of the content of images and videos
  • gesture recognition and handwriting
  • intelligent image processing.

Standardization in AI

Main article: Standardization of artificial intelligence

The impact of artificial intelligence

Impact on the economy and business

Impact on the labour market

Risks and challenges of using AI

Main article: Risks of using artificial intelligence

The development of artificial intelligence technologies in the future can carry not only benefits, but also harm.

Artificial Intelligence Technology Market

AI market in Russia

Main article: Artificial intelligence (Russian market)

Global AI Market

Scope of AI

The fields of application of AI are quite wide and cover both familiar technologies and emerging new directions, far from mass use, in other words, this is the entire range of solutions, from vacuum cleaners to space stations. All their diversity can be divided according to the criterion of key points of development.

AI is not a monolithic subject area. Moreover, some technological directions of AI appear as new sub-sectors of the economy and separate entities, while serving most areas in the economy.

The main commercial applications of artificial intelligence technologies

The development of the use of AI leads to the adaptation of technologies in classical sectors of the economy along the entire value chain and transforms them, leading to the algorithmization of almost all functionality, from logistics to company management.

AI in decision-making: Today and tomorrow

Main article: AI in decision-making: today and tomorrow

AI in public administration

Main article: Artificial Intelligence in Public Administration

The use of AI for defense and military purposes

AI in education

AI to solve demographic problems

AI in forensics

Main article Artificial intelligence in forensic science

AI in the judicial system

Main article: Artificial intelligence in the courts

AI in sports

AI in Medicine, Healthcare and Pharmaceuticals

Analysis of citizens' behavior

AI in housing and communal services

Machine Learning Tasks:

  • forecasting the technical condition of the house (elevator, roof)
  • prediction of water and electricity consumption (regression), prediction of parameter filling (classification)
  • count photo recognition

Artificial intelligence in software development

How AI helps to write software. Overview of one of the most promising technologies of the future

2024: Nvidia CEO: Don't teach kids programming - AI will replace developers

At the end of February 2024, Nvidia CEO Jensen Huang shared his vision of the prospects for the introduction of artificial intelligence in the field of software and services. In his opinion, there is no longer a need to study programming, since in the future AI will replace developers when writing code.

Automated software creation is considered one of the most promising areas of application of generative AI (Genia). Such tools will not only improve performance, but will also help improve the quality of program code and reduce the number of possible errors.

Jensen Huang shares his vision of the prospects for the introduction of artificial intelligence

According to Huang, as Genia systems improve, the need for programming specialists will rapidly decrease. And therefore, the head of Nvidia does not consider it necessary to teach children the skills of writing code - in the future these tasks will be able to fully take on AI. It will be enough for a person to formulate a task in a natural language.

File:Aquote1.png
Our task is to create a computing environment in which no one has to program. Everyone in the world will become programmers. It's an artificial intelligence miracle, "Huang says.
File:Aquote2.png

Against the backdrop of the rapid development of Genia, according to the CEO of Nvidia, people should focus on acquiring knowledge and skills in other important areas, such as biotechnology, education, production and agriculture. Huang's statements contradict the well-established view that "if a young man wants to succeed in the IT industry, he must learn to program." However, as of the end of February 2024, there are various AI-based programmer assistants that generate code in response to text requests and provide support in solving certain problems.[1]

Using AI in business

AI in the fight against fraud

On July 11, 2019, it became known that in just two years artificial intelligence and machine learning will be used to counter fraud three times more often than in July 2019. Such data were obtained during a joint study by SAS and the Association of Certified Fraud Examiners (ACFE). As of July 2019, such anti-fraud tools are already used by 13% of organizations that took part in the survey, and another 25% said they plan to implement them within the next year or two. Read more here.

AI in the electric power industry

  • At the design level: improved forecasting of generation and demand for energy resources, assessment of the reliability of power generating equipment, automation of increasing generation when demand jumps.
  • At the production level: optimizing preventive maintenance of equipment, increasing generation efficiency, reducing losses, preventing theft of energy resources.
  • At the promotion level: optimization of pricing depending on the time of day and dynamic charging.
  • At the level of service delivery: automatic selection of the most profitable supplier, detailed consumption statistics, automated customer service, optimization of energy consumption taking into account customer habits and behavior.

AI in the production sector

Main article: Artificial intelligence in the production sector

  • At the design level: improving the efficiency of new product development, automated evaluation of suppliers and analysis of requirements for spare parts and parts.
  • At the production level: improving the task execution process, automating assembly lines, reducing the number of errors, reducing the delivery time for raw materials.
  • At the promotion level: forecasting the volume of support and service services, pricing management.
  • At the level of service provision: improving the planning of fleet routes, demand for fleet resources, improving the quality of training of service engineers.

AI in banks

Front:

Middle/Back:

  • Detection froda
  • AML & KYC
  • Credit ratings
  • Risk Management
  • Compliance
  • Document processing

The main commercial areas of application of artificial intelligence technologies in banks

Investment AI

2024: One of the world's largest investment companies, Vanguard Group, implements $13 billion in AI to manage funds

In early February 2024, it became known that one of the world's largest investment companies, Vanguard Group, is introducing artificial intelligence technologies to manage several shareholder funds with a total capital of $13 billion. It is assumed that neural networks will help to adapt faster and more efficiently to changing economic and market conditions. Read more here.

AI on transport

Main article: Artificial intelligence in transport

AI in logistics

Main article: Artificial intelligence in logistics

AI in audit

Main article: Artificial intelligence in audit

AI in trade

AI in agriculture

AI in restaurants

2023: The first cafe, menus and interiors for which artificial intelligence has developed has opened in Russia

The first cafe, menus and interiors for which artificial intelligence has developed has opened in Russia. We are talking about an institution of Asian cuisine called Futuramen, which started working in Moscow on Pyatnitskaya. Read more here.

AI in jurisprudence

Main article: AI in jurisprudence

AI in the fashion industry

Main article: Artificial intelligence in the fashion industry

AI in science

Main article: Artificial intelligence in science

AI in the development of culture

Media and literature

How do robots replace journalists, writers and poets?

Video

Music

Main article: Artificial Intelligence and Music Creation

AI in painting

Main article: Artificial intelligence in painting

Games (go, poker, chess)

  • In the summer of 2017, it became known that Microsoft Research and Maluuba, a deep learning startup acquired by the corporation in early 2017, taught artificial intelligence to play one of the most popular computer games of all time, Ms. Pac-Man. And they did not just teach, but made him a champion who broke the world record set by man.

Playing the version of the famous arcade game M. Pac-Man, released for one of the first home consoles Atari 2600, artificial intelligence was able to score the maximum number of possible points - an achievement that was previously unthinkable. The result of the smart car was 999,990 points, while the best result set by a person is 266,360 points.

Artificial intelligence training used a method called "hybrid award architecture." It consists in the fact that 150 special agent programs are assigned a specific task: to avoid ghosts, move around correctly, collect pellets, and so on. With the help of agent programs, artificial intelligence independently allocated priorities to achieve maximum results. The Atari 2600 version of M. Pac-Man was used for a reason. The game code in it is less predictable than in the original version. The development strategy was the use of a promising approach to reinforcement learning (reinforcement learning), which assumes that the algorithm is given examples of the desired behavior for processing, and it is being improved by trial and error. According to scientists who worked on the project, such an achievement will contribute to the processing of natural language, and will also potentially form the basis of systems of detailed prediction of purchasing behavior due to many factors.

  • In 2016, the computer first beat a person[2]May 2017, the strongest go player Ke Jie from China lost the second game to the AlphaGo program . Thus, AlphaGo secured victory in a three-game tournament. Ke Jie, experts who followed the match noted, "perfectly" started the game, creating difficult combinations for the opponent throughout the playing field. However, AlphaGo managed to simplify the game and achieve victory.

  • In 2017, poker was threatened - specialists from Carnegie Mellon University created a bot that challenged professional players. The Libratus program, developed at Carnegie Mellon University, won the 20-day poker tournament "Brains Vs. Artificial Intelligence: Upping the Ante». The computer won more than $1.7 million in chips, according to New Scientist[3].

In the
tournament, which was held at the Pittsburgh casino Rivers, 120 thousand distributions were played in the unlimited Texas hold'em one-on-one (Heads-Up), Danielle McAulay, Jimmy Choo, Dong Kim and Jason Les played against Libratus. As a result of the 20-day tournament, the program defeated people, earning more than $1.7 million in chips. Despite this, the developer will not receive any money, and the prize pool of 200 thousand dollars will be divided between four live players, depending on the place occupied.

It is not known exactly how Libratus works, the authors described only the general structure of the program and plan to publish an article in a peer-reviewed journal in the near future. According to the developers, Libratus consists of three parts. The main "core" of Libratus was prepared in advance, calculations took 15 million core-hours, while Claudico took two to three million. The second part of the program monitored possible errors that opponents could make, and took this information into account during the game. The third part of Libratus tracked its own weaknesses that opponents could exploit, and adjusted the overall strategy with this data in mind. This approach allowed the program to both bluff on its own and recognize disinformation from rivals[4].

According to the authors of the program, systems like Libratus have a great future in various areas where you have to deal with incomplete information. Researchers call information security, military affairs, auctions, negotiations and even the lean distribution of medicines as possible areas of application of the program.


Poker is a game that is very difficult to train to play a computer: a good player quickly recognizes strategies embedded in artificial intelligence and finds a way to defeat the bot. It is especially difficult for a computer if bets at the poker table are not imitated, that is, the player can set an unlimited number of chips in his turn.

However, poker bots are a very popular trend for the game. There are two types of poker bots. Some are quite simple and fight people in a game with small stakes - in it, the level of poker is very low, and people cannot solve even the simplest strategies. Such bots are not very interesting to science and serve to make money - poker sites, as a rule, try to fight them.

The second type is bots that compete with professionals. They are needed not only and not so much to make money, but to promote science. The topic of "games with incomplete information" is now one of the most popular in economic science - it is no coincidence that Lloyd Shapley and Alvin Roth received the Nobel Prize in Economics in 2012 precisely for the theory of stable distribution, which is connected precisely with "game theory." If a computer consistently learns better than a person to play games with incomplete information, we may no longer have to bargain and agonize about whether we lost the game by buying a new car with the characteristics we need for this particular price - because it will be up to us to decide on an application in a smartphone[5].

  • Computer game developers use AI to one degree or another. This forms the concept of "Gaming Artificial Intelligence." The standard tasks of AI in games are finding a path in two-dimensional or three-dimensional space, imitating the behavior of a combat unit, calculating the correct economic strategy, and so on.

Photo

Main article: Artificial intelligence in photography

Chips for artificial intelligence

Main article: Chips for artificial intelligence

Read also

Listen also

Robotics



  1. Nvidia CEO Jensen Huang's message to kids: generative AI means you don't need to learn coding
  2. in Go-pro Artificial intelligence for the first time defeated a professional Go player: Wired The Go game was invented more than two and a half thousand years ago and is still one of the most popular games in the world - championships are regularly held on it. At first glance, it is very simple: there is a field of cages and stones - black and white. Players must capture as much area as possible on the board with their stones. Nevertheless, it was this game that was beyond the control of a computer for many years. Until recently, artificial intelligence could not beat high-level players - masters.. In
  3. AI just won a poker tournament against professional players
  4. Artificial intelligence defeated professional poker players
  5. Computer against a person. This time in poker, Artificial Intelligence is trying to beat four professional players