The world of technology is already an essential part of our daily lives, and it is interesting to understand the characteristics of artificial intelligence that have made a difference in the way we live. Did you know that artificial intelligence is a branch of Computer Science? Enter here, and learn about its features, and much more.
Table of Contents
- 1 Features of artificial intelligence
- 2 What is artificial intelligence?
- 3 Artificial intelligence classification
- 4 Topics in artificial intelligence
- 5 Applications of artificial intelligence
- 6 Conclusions
Features of artificial intelligence
In this section we will discuss the characteristics of artificial intelligence. In this case, artificial intelligence (IA) is developed in one of the branches of Computer Science, where logical algorithms are applied that seek to imitate the cognitive behavior of the human brain. Of course, the definition of artificial intelligence can continue to develop, but in the end all the characteristics of artificial intelligence will agree that they are used for the programming of robotic devices.
In the summer of 1956, the Dartmouth Conference was held to address the issue of artificial intelligence, in which John McCarthy, Marvin Minsky and Claude Shannon participated. This meeting is when the term artificial intelligence was first implemented, where they stipulated some forecasts to ten years which were not met, so the investigations were abandoned for approximately fifteen years. It should be noted that the term "artificial intelligence" is attributed to John McCarthy.
It's easy to think that it's only a matter of time before technology and artificial intelligence completely replace humans. In fact, there are movies and real-life researchers who believe that machines with artificial intelligence will be capable of subjugating the human race and enslaving it. At the moment, this is very far from reality, because this will only be possible when the artificial intelligence has consciousness and has the ability to create a new device with artificial intelligence on its own, and manages to bypass and override the commands of its programming of its own volition. At that point, the human being would lose control of the situation.
What is artificial intelligence?
It was precisely during the Darthmouse conference in 1956, where the term artificial intelligence was officially defined, which establishes that if a machine or a robot behaved in a similar way to the behavior that a human being would perform, then it would be considered as a device. smart.
Other definitions attributed to artificial intelligence are the following:
act like people
This is the definition established by McCarthy, which refers to evaluating the behavior of the machine to determine if it can be considered intelligent. The so-called "Turing Test" applies this definition to define the results of its Test. All devices that are similar in actions such as decision making, problem solving and learning as humans would do, meet the characteristics of artificial intelligence.
The test proposed by Alan Turing is that a human being will carry out a conversation in natural language with a machine and a human being at the same time, the machine would seek to imitate the behavior of a human being and would try to deceive its evaluator through its answers to make him believe that he is actually a human being. In other words, between The characteristics of artificial intelligence is the ability to imitate humanity.
Of course, the tester must know in advance that he is talking to a machine and to a human and must try to determine which is the real human and which is the impostor.
In this case, the ability to speak would not be taken into account, since the evaluator would be placed in a separate room where he would receive the information through a computer, so communication would depend entirely on the keyboard and not on this ability. In this sense, between The characteristics of artificial intelligence is to simulate the human voice.
reason like people
Another of the characteristics of artificial intelligence is in the definition of assessing the development of the rationing made by the robot, without taking into account whether the result obtained was successful or not. This view is used by cognitive science. In this reasoning, all the calculations necessary to be able to perceive, reason and act against the event are executed.
Like the previous definition, one of the characteristics of artificial intelligence is the rationing done by the machine, however, it takes into account if this rationing has a logic and coherence, so that said rationing has been carried out.
In this point of view if the results are considered again. Using the chess-playing robot as an example, its goal is to win every game. Another feature of artificial intelligence is the ability to perform calculations, which will be indifferent as long as it reaches the goal.
Artificial intelligence classification
Artificial intelligence can be classified according to its objectives.
weak artificial intelligence
This point of view considers that computers can only pretend that they have rationing, and not that they actually have one of their own. The researchers who support this position consider that it is not possible for a computer capable of consciousness to exist or be developed, since in reality it would be a program that would simulate such a thing.
strong artificial intelligence
On the other hand, there are researchers who affirm the fact that a computer can have reasoning or thoughts with the same capacities as the human mind. This would mean that a computer would be able to reason, imagine, feel, among other things, on its own, even when everything starts from a program, its neural network could evolve until it reaches this point.
Topics in artificial intelligence
Although the definitions and points of view regarding artificial intelligence all converge on four issues to consider to attribute to a device machine if it has artificial intelligence.
Troubleshooting and Search
One of the main objectives of artificial intelligence is to solve the problems for which they are designed. In the first place, when posing a problem, it is necessary to formalize it in a way that then allows its solution. This topic focuses on the search for the formalization of problems and their resolution.
Knowledge representation and knowledge-based systems
This topic focuses on those problems that require previous knowledge to be able to solve them. For example, those artificial intelligence programs that are applied in medicine, it is necessary to incorporate knowledge and information regarding the subject so that it can solve the problems of this subject.
This topic refers to the learning process carried out by the machine according to the experiences obtained. There are different types of learning such as imitation learning, reinforcement learning, deep learning or decision tree based learning. All these types of learning allow the machine to store the actions carried out that considered the final objective fulfilled, in order to apply the same actions in the event of a similar event.
Reinforcement learning is the same as that used to train animals, that is, when they perform a task or correctly obey an order, they receive a reward. In this case, the machine receives its first order and as it gets positive results, it receives this as an incentive to continue improving its decision making. For example, depending on the Types of robots You can consider winning the game of chess as your prize.
Another type of learning is called deep learning, in which the imitation or similar behavior of the neural network and communication processes that occur in the neurons of the human brain are sought.
For example, when the natural sensors of the human body such as the eyes, ears, touch, taste or smell detect a variation, a signal is sent to the brain. This signal is received and analyzed by a first neuron that communicates the detection of a change to the following neurons and thus initiates a neuronal sequence to understand the event and how to react.
A similar process occurs when, for example, facial recognition cameras detect an individual through their visual sensors, it is activated. When detecting the face, it starts a sequence of logical processes starting from the analysis of the simplest data such as the colors that face has. Then, it seeks to determine the geometric figures that make up that face. Finally, divide the face into multiple frames to better define the details that distinguish that face.
This type of learning uses different problem-solving schemes that are activated as information is received. If the example of the robot that plays chess is taken up again, it will start its scheme in which is the first piece that its opponent moved, there it will carry out multiple calculations corresponding to the statistics of which one it should move. Later, his opponent will make another move and a new scheme will open where he will again make calculations in order to make his next moves.
Finally, when it succeeds in winning the chess game, then the robot stores all the decisions that it and its opponent made for future chess games, so that when a similar event happens, it already has the necessary information in the database. and can respond more quickly and with a higher percentage chance of winning the game.
distributed artificial intelligence
Thanks to the advances that have allowed us to know How technology works, such as the evolution of multiprocessors and the appearance of the internet, have allowed artificial intelligence to provide distributed solutions.
Applications of artificial intelligence
Additionally, there are four branches that are closely linked to the use of artificial intelligence, which are:
- Natural language.
- Artificial vision.
- The robotic.
- speech recognition.
Currently, various applications that use certain algorithms or methods have been developed in the area of artificial intelligence.
Even when mention can be made of many applications of artificial intelligence, it would be difficult to cover all those in which its presence is found, since today there are devices for everyday use, or programs used by companies and corporations where find this technology.
For example, today a supercomputer is being used that is applying an algorithm which makes combinations of different drugs in order to try to find a cure for Covid-19. This computer evaluates the symptomatology data, the composition of the virus and other information necessary to be able to counteract it, and through the database that contains the different drugs that exist, it makes combinations trying to cure the patient of this disease, taking into account even the side effects that these combinations can cause and the recommended doses.
Another example, may be those used by multiple search engines that use the learning method to know the interests of each user individually, this allows you to create behavioral profiles and preferences for and thus be able to provide ads according to these. pleasures.
Next, we will present some of the most outstanding applications of artificial intelligence.
Applications in games
There is a robot with the ability to beat even the best players in chess, since this robot was built for the sole purpose of performing the necessary calculations and statistics to establish strategies in their moves and win each game.
Today practically all games have managed to be beaten by a machine, although the first table games to be beaten by a machine that had artificial intelligence were checkers and Othello.
The University of Alberta in 1989 developed a program called Chinnok by Jonathan Schaeffer's team and it was in 1994 that he became the world champion in ladies. The Chinnok program has a database of checkers game openings and closings made by the best checkers players.
Again, in 2007 it was shown that when the game is done perfectly, it is impossible to program Chinnok. And when a match is played with an improvement in the strategy by the opponent, at most it can reach a draw against this program.
In the case of chess, different innovations and problem-solving programs have been developed to win this game for years, however it was in 1997 in the month of May in New York when Deep Blue defeated the world champion G. Kasparov .
It was a software developed by the IBM company that had specific hardware, and databases that made it possible for this program to culminate perfectly when the final situations were presented with seven or even fewer pieces on the board. Likewise, its search algorithms, the minimax type, were able to determine the best options in all the different cases.
Today it is the public game in which a machine with artificial intelligence to beat a human player, however this is not surprising, since for some time Go has been considered an even more difficult and complex game than chess. .
In addition, the dimensions of this game also considerably increase the difficulty since it has more than 361 intersections to make a board of 19 3 19, without mentioning the number of possible moves that can be made on each board.
Even though there has not been a machine capable of winning this game, there are already programs that respond well to boards with dimensions of nine by nine, and unlike the search algorithms used for the game of chess, in this case search algorithms are used. UCT search.
Robots have various areas in which they provide their support for faster, more efficient and precise solutions, such as in production lines that require process automation, also in the military and defense fields, and even for exploration. space as is the case of the Curiosity mobile robot that is currently on Mars, in order to collect information regarding the possible existence of life on this planet.
Today, there are robots that serve as entertainment and participate in games, such as Japanese pet robots called Paro and Aibo. In the case of Paro, it is a therapeutic robot that helps reduce stress levels in patients and helps improve their socialization. In the case of Aibo, it is a dog-shaped robot that was developed by Sony, which has a vision system and is programmable.
exploration and reconnaissance robots
There are mobile robots that are used for exploration, search and reconnaissance in hostile environments or areas. For example, like the robots used in the Chernobyl nuclear disaster that tried to visualize the damage caused by the incident and managed to capture images of the radioactive mass called the Elephant's Foot.
Or we can also mention the robots Spirit, Opportunity and Curiosity that were sent to the surface of the planet Mars, in 2004 the first two and in 2012 the third, which fulfill the mission of analyzing all biological, atmospheric and biological processes. components that make up this planet.
In 1997, the Honda company presented the first bipedal robot, that is, it had the ability to walk with two limbs and was called P3. Again, Honda introduced in 2000 the ASIMO robot that comes from the diminutive Advance Step in Innovative Mobility. This was the end of the Honda P robot series. All of these robots were purposely designed to have the physical structure and motor capabilities of a human being.
Now, ASIMO can change whether it is running, climbing stairs and avoiding obstacles, and even through its visual sensors or cameras it can recognize moving objects, gestures and postures.
Smart vehicle applications
One of the most recent innovations are autonomous passenger vehicles.
The first metro with fully automated driving emerged in the Japanese city of Sendai, which was developed in 1987. Today there are already many fully automated metro systems.
Another example of vehicles that can carry passengers and can be fully automated is the Stanley which was the winner in the 2005 DARPA Challenge competition, which took place in the Mojave desert. Stanly achieved which he managed to successfully complete the 212,4 kilometer route in a time of six hours and 54 minutes.
Later, in the 2007 DARPA Grand Challenge held at George Air Force Base, the Stanley automated vehicle again successfully completed the 96-mile course. The vehicles that participated in this race were capable of processing the traffic rules of the state of California in real time.
In another part of the world, specifically between the International Congress Center and the Brandenburg Gate, the Made in Germany vehicle developed by the Free University of Berlin traveled on a 80-kilometre route. This vehicle is fully autonomous, it has the ability to recognize the presence of pedestrians and traffic lights. However, it requires data such as travel speed to be provided.
unmanned aerial vehicles
Also known as UAV from the diminutive of Unmanned Aerial Vehicle. The first unmanned aerial vehicle to cross the Pacific Ocean without the need to make a stop was the Global Hwak. This was carried out in 2001 in the month of April, it began in the United States and ended in Australia.
This model, however, still relies on a ground station pilot and other operators for data analysis. In fact, Weiss indicated in 2011 that these systems, even though they are capable of collecting large amounts of information, still do not have the necessary capacity to process it in real time, and therefore respond to events intelligently according to the information. instantly collected.
These vehicles today are more popularly known as Drones. Drones have various internal sensors and devices that help you in your navigation. For example, they have a GPS module for geolocation, cameras with wireless connection, some with motion and heat sensors, among others. In the first instance, this technology arose for military use, although it is already on the market.
There is no doubt that various areas of technology have advanced exponentially and computing does not escape this advance, in fact it has rather fostered the evolution of other branches of science. The calculation capacity provided by artificial intelligence has been doubling in a period of eighteen months according to Moore's Law.
This would imply that if Moore's Law continues to hold, then by around 2030 the computing power of processors will be similar or perhaps equal to that of a human being.
Search engines like Google and Amazon store millions of pieces of information from their users in order to determine the preferences of each individual in order to provide a better service. So servers with large memory capacities have been required to record evenly in this data.
Social networks, likewise, require these large storage capacities to record the preferences of their consumers in order to present proposals according to their tastes.