Artificial Intelligence, abbreviated as AI, first appeared in our spoken language in 1889. It is also called machine intelligence and refers to the intelligence displayed by machines made by humans. Generally speaking, artificial intelligence refers to technology that uses ordinary computer programs to represent human intelligence. The term also refers to the study of whether and how such intelligent systems can be implemented. At the same time, through advances in medicine, neuroscience, robotics, statistics, etc., normal predictions believe that many human occupations will gradually be replaced by them.
The field of artificial intelligence, defined in general textbooks, is “the research and design of intelligent agents.” An intelligent agent refers to a system that can observe the surrounding environment and take action to achieve its goals. John McCarthy defined 1955 as “the science and engineering of making intelligent machines.” Andreas Kaplan and Michael Heinlein define artificial intelligence as “systems that correctly interpret external data, learn from this data, and use this knowledge to achieve specific goals and tasks through flexible adaptation ability.”
Artificial intelligence is a machine or computer that imitates human cognitive functions associated with human thinking, such as learning and problem-solving. Artificial intelligence is a branch of computer science that senses its environment. It takes actions to maximize its chances of success. In addition, artificial intelligence can learn from past experiences, make reasonable decisions, and respond quickly.
Therefore, the scientific goal of artificial intelligence researchers is to understand intelligence by constructing computer programs that embody reasoning or reasoning. The four main components of artificial intelligence are:
Acts as an expert in handling the situation under review and produces expected or expected performance.
Heuristics
It involves evaluating a small range of solutions and may include some guesswork to find a solution close to optimal.
Natural language processing: realizing communication between humans and machines in natural language.
Computer Vision
Automatically generates the ability to recognize shapes and functions.
Research regarding artificial intelligence
The research on artificial intelligence is highly technical and professional. Each branch field is in-depth and interconnected, so it covers a very wide range. Research on artificial intelligence can be divided into several technical issues. Its subfields focus on solving specific problems, one of which is how to use a variety of different tools to complete a specific application.
The core issues of AI include the ability to construct reasoning, knowledge, planning, learning, communication, perception, movement, moving objects, using tools, and controlling machinery that is similar to or even superior to humans. General artificial intelligence (GAI) is still the long-term goal of this field. At present, weak artificial intelligence has achieved preliminary results. It has even reached a level beyond human capabilities in image recognition, language analysis, board games, etc. The versatility of artificial intelligence means that what can solve the above problems is the same AI program can directly use existing AI to complete tasks without re-developing algorithms. It has the same processing power as humans.
However, it will take time to research and achieve integrated artificial solid intelligence with the ability to think. Popular methods include statistical methods. Computational intelligence and AI in the traditional sense. There are currently a large number of tools that apply artificial intelligence, including search and mathematical optimization, as well as logical deduction. Algorithms based on bionics, cognitive psychology, probability theory, and economics are also being gradually explored.
Introduction of Artificial Intelligence (AI) into two parts
The definition of artificial intelligence can be divided into two parts: “AI” and “intelligence.” “Artificial” means designed by people, created by people, and manufactured for people.
The controversies about the word intelligence
There is some controversy about what “intelligence” is. This involves other issues such as consciousness, self, mind, including the unconscious spirit, etc. It is a generally accepted view that the only intelligence that people understand is their intelligence. However, our understanding of our intelligence is minimal, and our knowledge of the necessary elements constituting human intelligence is also minimal, so it is difficult to define what “artificially” manufactured “intelligence” is.
Therefore, research on artificial intelligence often involves the study of human intelligence itself. Other intelligence related to animals or other artificial systems is also generally considered a research topic related to artificial intelligence.
Current use of artificial intelligence
Artificial intelligence is currently being used more and more extensively in the computer field. It is applied in robots, economic and political decision-making, control systems, and simulation systems. Artificial intelligence is also widely used in many different fields. Robots run restaurants and shops and repair city infrastructure. Artificial intelligence manages transportation systems and autonomous vehicles.
Innovative platforms manage multiple urban areas, such as waste collection and air quality monitoring. Urban AI is embodied in urban spaces, infrastructure, and technologies, making our cities unsupervised autonomous entities. Digitally enabled intelligent response services can be easily implemented in real-time. Many cities are now proactively leveraging big data and artificial intelligence to improve economic returns by providing better energy, computing power, and connectivity to our infrastructure.
AI for various public services
Recently, many governments have started using AI for various public services as it reduces administrative costs and time. For example, robotic automation of immigration processes reduces processing times and increases efficiency. Artificial intelligence brings technological breakthroughs to local government services. Artificial intelligence agents assist urban planners in scenario planning based on goal-directed Monte Carlo tree search. Targeted reasoning AI agents provide optimal land use solutions and help us develop democratic urban land use planning.
Artificial intelligence leverages online data to monitor and modify environmental threat policies. During the 2019 water crisis, the Latent Dirichlet Allocation Method identified the most discussed topics on Twitter (X), a naive method of categorizing tweets on the effects and causes of drought, government responses, potential solutions, etc. Topics are categorized. AI tools complement human judges in the judiciary to provide objective and consistent risk assessments.
Development history and sub-fields of (AI)
The direction of artificial intelligence research has been divided into several sub-fields. Researchers hope that an artificial intelligence system should have certain specific capabilities. These capabilities are listed and explained below.
Reasoning and problem-solving
Early AI researchers directly imitated humans’ step-by-step reasoning, much like how humans think when playing board games or performing logical reasoning. In the 1980s and 1990s, artificial intelligence research also developed very successful methods for dealing with uncertain or incomplete information using concepts from probability and economics.
difficult problems
For complex problems, a large amount of computing resources may be required; that is, a “possible combinatorial explosion” occurs: when the problem exceeds a particular scale, the computer will require astronomical amounts of memory or computing time. Finding more efficient algorithms is a priority for artificial intelligence research projects.
Sub-representational problem-solving method
The human problem-solving model usually uses the quickest, intuitive judgment rather than conscious, step-by-step derivation. Early artificial intelligence research usually uses step-by-step derivation. Artificial intelligence research has progressed in this “sub-representational” problem-solving method: research on embodied agents emphasizes the importance of perceptual movement. Neural network research attempts to recreate this skill by simulating the brain structure of humans and animals.
Knowledge representation and common sense knowledge base artificial intelligence
Knowledge representation is one of the core research issues in artificial intelligence. Its goal is to allow the machine to store corresponding knowledge and be able to reason and deduce new knowledge according to specific rules. Many problems that require a large amount of knowledge about the world need to be solved, including a priori knowledge stored in advance and knowledge obtained through intelligent reasoning. Prior knowledge stored in advance refers to the knowledge that humans tell the machine in some way.
Knowledge obtained through intelligent reasoning refers to knowledge obtained by combining prior knowledge and certain specific reasoning rules (logical reasoning). First, prior knowledge can refer to knowledge that describes goals, characteristics, categories, and relationships between objects. It can also describe events, time, states, causes and results, and any knowledge you want the machine to store.
Example
For example, there is no sun Today. If there is no sun, it is cloudy. Then, in propositional logic language, this knowledge can be expressed as Today → there is no sun, there is no sun → it is cloudy. This knowledge is a priori knowledge so that reasoning can obtain new knowledge: Today → cloudy day. From this example, we can see that the correctness of prior knowledge is essential. In this example, if there is no sun, it is cloudy. This proposition is not rigorous and relatively general because it may rain or snow without sun.
In addition, if artificial intelligence can see the sun, in addition to the question of how to judge, under this premise, it should also be able to judge the difference between cloudy and sunny days. Logical proposition representation is significant in knowledge representation, and logical inference rules are currently the main inference rules. You can use logical symbols to define each logical proposition in the machine and then let the machine store the corresponding logical reasoning rules so the machine can naturally perform reasoning. Many difficulties in knowledge expression cannot be solved.
For example, it is almost impossible to build a complete knowledge base, so the resources of the knowledge base are limited; the correctness of prior knowledge needs to be tested, and previous knowledge is sometimes not necessarily only Two choices, right or wrong.
planning
An intelligent agent must be able to set goals and achieve those goals. They need a way to build a predictable world model (representing the entire world state as a mathematical model and predicting how their actions will change the world) to choose the most effective actions. In traditional planning problems, the intelligent agent is assumed to be the only influential one in the world, so its behavior has been determined.
However, if this is not the case, it must periodically check whether the state of the world model is consistent with its predictions. If not, it must change its plans. Therefore, intelligent agents must be able to reason in uncertain states. In a multi-agent system, multiple agents plan to accomplish specific goals cooperatively and competitively. Using evolutionary algorithms and group intelligence can achieve an overall emergent behavioral goal.
Machine learning
The primary purpose of machine learning is to allow the machine to obtain knowledge from users and input data so that the machine can automatically judge and output corresponding results. This approach can help solve more problems, reduce errors, and improve problem-solving efficiency. For artificial intelligence, machine learning has been meaningful from the beginning. 1956, at the original Dartmouth Summer Meeting, Raymond Solomonov wrote about probabilistic machine learning without supervision: a machine for inductive reasoning.
Two categories of machine learning ( supervised learning and unsupervised learning)
There are various methods of machine learning, which are mainly divided into two categories: supervised learning and unsupervised learning. Supervised learning refers to giving the machine some training samples in advance and telling the category of the sample, and then training based on the category of these samples to extract the common attributes of these samples or train a classifier.
When a new sample comes, the common attributes obtained through training will be Attributes or classifiers used to determine the sample category. Supervised learning is divided into two categories: classification and regression based on the discreteness and continuity of the output results. Unsupervised learning does not give training samples but gives some samples and rules, allowing the machine to automatically classify according to some rules. No matter which learning method is used, error analysis will be performed to know whether the proposed method has an upper limit on the error in theory.
Natural Language Processing
This processing explores how to process and use natural language, while natural language cognition refers to making computers “understand” human language. Natural language generation systems convert computer data into natural language, and natural language understanding systems convert natural language into a form that is easier for computer programs to process.
