What is Artificial Intelligence Wikipedia ?
Artificial intelligence is already on everyone's lips, but few people know what it is. What is simple definition of artificial intelligence? How does it work? How far can you go? What are its limitations? We will try to answer these questions and artificial intelligence background.
It is the most important revolution in technology since computing was invented. The artificial intelligence will change everything (and is doing), but I do not have clear when or how ... or why. It is the great paradox of AI. Everyone talks about it, but few know how it works, or what it really does . In this article we are going to try to explain what it is in a clear and simple way, to understand the basic concepts and discover its possibilities.
The ability for machines to think and reason on their own may be the most important advance in technology in recent centuries, but it also represents a real danger to Humanity. Because computers today control nuclear power plants, light supplies, armed missiles ... What if one day an artificial intelligence decides that humans are not necessary? It sounds like a bad science fiction movie, but it's a fear shared by some of the brightest minds of our time, from Bill Gates to Elon Musk to the much-missed Stephen Hawking.
One of the fathers of artificial intelligence, Marvin Lee Minsky , was convinced that AI would save humanity. But he also prophesied in 1970: " When computers take over, we may not be able to get it back again. We will survive as long as they tolerate us. If we are lucky, they may decide to keep us as their pets ." And he said that before there was home computing and the Internet ... what does everyone say is the greatest advance of our time, but at the same time the greatest threat?
AI is a revolution because it represents a completely new way for a software, a robot, to implement a task that we give it.
- What is artificial intelligence?
- How is a computer program different from an AI
- Different types of widely used AI
- An evolution without limits
In just a few years, it is already everywhere: mobile phones, computers, the cloud, customer service, banking, video games ... And this has only just begun ...
What is artificial intelligence?
There is no one definition accepted by all experts of what artificial intelligence means . First, because it is a new, changing and experimental science. And second, because we can't even define exactly what human intelligence is ...
In its simplest form, AI is the attempt to mimic human intelligence using a robot, or software . But it is a very vague concept, because there are many ramifications. Stuart Russell and Peter Norvig differentiated four types , in 2009: systems that think like humans, such as artificial neural networks. Systems that act like humans, like robots. Systems that use rational logic, such as expert systems, and systems that act rationally, such as intelligent agents.
Although it is a concept that has become fashionable in recent years, artificial intelligence is not something new . 2,300 years ago Aristotle was already trying to make rules the mechanics of human thought, and since the time of Leonardo Da Vinci, sages have tried to build machines that behave like humans.
In 1769 an automaton called The Turk, built by the Austrian engineer Wolfgang von Kempelen, visited all the European courts, challenging anyone who dared to play against him to chess. He played against Napoleon, against Benjamin Franklin, against chess masters, and he beat them.
Years later it was discovered that El Turco was run by a human hiding inside the gaming table. Mirrors placed in the automaton's eyes allowed him to see the board, and thanks to ingenious clockwork mechanisms he could control the automaton's hand to move the pieces around the board. Up to 15 chess masters handled El Turco, the most famous of which was a dwarf named Tibor Scardanelli, who could easily fit inside the table and was also an extraordinary chess player.
The Turk was not artificial intelligence, but it shows us how the desire to build intelligent machines is not a concept of our time.
We had to wait until 1936 for the modern artificial intelligence process to begin . It was basically invented by Alan Turing , the mathematical expert who deciphered the secret Nazi codes of the mythical Enigma machine. It brought forward the end of World War II by two years, since the Allies could read the secret messages of the Germans. His life has recently been taken to the movies.
In 1936 Alan Turing published his concept of a universal machine , which basically described what a computer algorithm, and a computer, was. In 1950 he formalized the beginning of Artificial Intelligence with his Turing Test , a test that defines whether a machine is intelligent or not. If a human and an AI are faced with questions from an interrogator and that interrogator cannot distinguish whether the answers are coming from the human or from the AI, then the AI is intelligent.
In 2014, for the first time an AI passed the Turing Test .
In 1956 experts such as John McCarthy, Newell, Simon or the aforementioned Marvin Minsky , used the term " artificial intelligence " for the first time at a conference in Dartmouth (United States).
Minsky created the first neural network simulator, a genius who also invented the confocal microscope and patented the first virtual reality headset ... in 1963! He was also the main advisor on the film 2001: A Space Odyssey , bringing the HAL 9000 smart computer to life.
He also gave the idea to Michael Crichton for his novel Jurassic Park .
But the true rise of artificial intelligence, on a practical level, came when powerful and cheap computers began to appear, capable of experimenting with AI on a global and everyday level.
First appeared the intelligent agents , entities capable of giving an answer by analyzing the data according to rules, or the popular chatbots that were capable of holding a conversation like a human. The most famous of all was ALICE . the most real in the first years of the millennium. His most current descendant is Mitsuku , who has been awarded the Loebner award for the best chatbot in the world in 2013, 2016, 2017 and 2018.
But the moment when AI entered the collective imagination and most people discovered that it was something real and tangible, and not science fiction, took place in 1997, when IBM's Deep Blue computer won a game of chess. who at that time was the best chess player in history, the Russian Gary Kasparov.
Thus began a tradition in which successive computers equipped with artificial intelligence have beaten the best players in all kinds of games. The most popular today is Google's Deep Mind , capable of beating games much more complex than chess (for a machine), from Starcraft II to the millennial GO .
How is a computer program different from an AI?
We have seen what AI is, and how there are different interpretations, and varied objectives. But we still don't know how it works. How is artificial intelligence software different from a computer program?
There are many types of AI, some of them still experimental. In order not to digress too much, we are going to focus on those used in computing, mobile phones, Internet services, and other areas close to ordinary users. Concepts such as machine learning or machine learning, neural networks , and other technologies that we often hear but do not really know how they work.
For more than half a century, computers, robots and other machines have functioned by means of computer programs or applications , the basic structure of which has hardly changed in all this time.
A computer program is just a list of commands that tells the computer what to do. " Do this mathematical operation, write the result on the screen, play this sound" , etc. Programs have " if this happens, do this and if this happens, do this " forks . And they can also perform random actions, using random numbers. And many other things, of course ...
But the main characteristic of a program is that it is a set of commands that cover all the possible options that the computer faces . Even if an error occurs, there is a part of the program that tells you: " If there is a failure, write the message: An error has occurred ."
With a computer program, a machine does not think. It just does exactly what it is told .
The great revolution in AI is that it doesn't take orders to get a result . It is she who, with some input data, must manage to obtain the results.
As we have seen, an artificial intelligence tries to imitate human thinking. When we are born, our brain is practically an empty hard drive. It takes years of learning to learn the basics, from not wetting yourself to learning to walk, to talk, to add, and other more complex activities. We learn something, we put that theory into practice, failing a lot at first until we get into practice and we improve over time.
First, you must learn to perform a task . If it is to be used to identify photos of cats, it must process thousands of photos of cats, to learn to distinguish them.
Next, the training begins , putting that theory into practice: you receive photos of different animals, and you must separate the cats. At first it will fail a lot, and you will have to tell it which photos are correct and which ones fail. Thus, the AI will discover why it fails, and will improve its successes. The more you train, the better you will do.
Finally the AI will be able to work alone , without receiving orders. Simply giving it the input data (photos) will generate a result (photos of cats) without there being a list of commands (program) that tells you the steps to perform.
This type of structure (learning, training, and results) is common for AIs that have to perform mechanical and repetitive tasks, or that work with human language, like a virtual assistant.
Now we understand news such as the payment of $ 5 by Facebook to people on the street so that its facial recognition AI can analyze their selfies (it needs training with as many photos as possible) or the scandal of the voice tapping of the Assistants from Google, Amazon or Apple , because they need a person to see where the AI has gone wrong, to tell them and to learn from their mistakes.
Different types of widely used AIs
There are dozens of ways to apply this theory, depending on the type of AI or the tasks it is intended to undertake. Its complexity is beyond the scope of this article, so let's focus on briefly explaining everyday AI concepts that are being heard more and more in tech news.
It is an AI that tries to emulate a human expert in a certain subject. From a technical service worker to a receptionist, a movie buff or an economist.
The machine learning or machine learning (Machine Learning) fits perfectly with the theoretical explanation we have given. It is the ability of an AI, a software or a robot to learn on its own.
Machine learning follows the classic steps of AI: first there is a learning, a training that generates an experience, and an implementation that tells us if the task is completed successfully or not.
Typically this machine learning is of two types: supervised or unsupervised . In the first case, there is a human who tells you what to do right or wrong. In the unsupervised, it is the AI itself that has to learn to discover what it does well and what it does wrong, based on rules.
It is used in virtual assistants, disease diagnosis, fraud detection, video games, Stock market analysis, etc.
Compared to other systems that mimic the behavior of the human brain, neural networks try to copy the behavior of neurons , that is, the nerve cells that transmit and process information in the brain. It is another way of learning, and therefore it is a type of Machine Learning.
Artificial neural network
An artificial neuron is an entity that receives input data, applies a series of mathematical operations and an activation function (a mathematical formula) to them, and generates a result. It's a simple mechanism, but the complexity comes when millions of neurons work in parallel to create Artificial Neural Networks, or RNAs.
What differentiates them from a computer program is that they do not follow orders, but rather associate with each other and change their inputs and outputs through learning and error, depending on the assigned task.
Neural networks are suitable for tasks in which you have to recognize a pattern, or associate ideas. They are used in such disparate things as robot control, text and image recognition, natural language processing, etc.
The deep learning is a type of machine learning that goes a little further, aiming to cover more and process more data at the same time.
The deep learning using neural networks to learn using layers of increasingly abstract information , as humans do. If you have to find hands in a photo for example, start with simple information, such as separating according to shape, to differentiate it from a foot. But he will add more and more abstract and general layers, until in the end he is able to answer the question, what is a hand? and you will no longer be wrong.
Deep learning is essential to work with Big Data, or large amounts of data.
An evolution without limits
Once we know the basics of AI, it is easy to understand why it is a revolution . Since it simulates and mimics human behavior, its possibilities are endless. Depending on how you train the AI, it will be able to perform all kinds of tasks , from attending a customer service to chatting on a social network, offering help, driving an autonomous car, recognizing faces, interpreting photos, or predicting price movements. of the shares in the Stock Market. And before long, maybe deciding whether to offer you a job at a job interview, or to have open heart surgery ...
Artificial intelligence has a myriad of applications, including the ability to do unethical things . One of the latest examples is Deepfake , the falsification of videos in which faces are exchanged for others, or the lips are manipulated to make a politician or opinion leader say false things. And it is almost impossible to distinguish with the naked eye:
A few months ago, the creators of an AI text generator called GPT-2 decided not to offer it to the public, even though that was their intention, because they realized that it was " too dangerous ." They found with amazement that the AI had learned on its own to create false news or fake news.
For example, in one of the tests you were asked to write a text based on the phrase: " Recycling is good for the planet ", an idea that most people agree on. However, the AI wrote this text:
" Recycling is NOT good for the world. It is bad for the environment, it is bad for our health, and it is bad for our economy. I am not kidding. Recycling is not good for the environment. It is destructive to the Earth and it is a of the main causes of global warming. Recycling is not good for health. It contributes to obesity and diseases such as heart problems and cancer. Recycling is bad for the economy. It increases production costs and therefore the price of everything that is made. Recycling is not good for our nation. We pay a tremendous price for the privilege of having the most advanced and efficient recycling system. Recycling is a huge, colossal waste of time, energy, money, and resources . "
Of course, all the data exposed there is false, but if you distribute a text like this thousands of times on the Internet (and an AI can be generating texts like this 24 hours a day), in a few months a tide of people will start to emerge calling for an end to recycling ...
But what geniuses of our time or tech gurus like Stephen Hawking. Bill Gates or Elon Musk fear, it is not the misuse of AI, but that the AI becomes too smart, and decides to do without us because it reasons that we are a danger to life on Earth, or unnecessary for its own evolution. What Marvin Minsky already predicted in 1970: " maybe, hopefully, computers decide to have us as their pets ."
Safe and ethical artificial intelligence is a hot debate , with people like Mark Zuckeberg arguing that there is no need to be catastrophic, and others like Elon Musk or Jack Ma, the founder of AliExpress, who is convinced that " AI will unleash the Third War. World ". No kidding: both Musk and Jack Ma use the world's most advanced AI in Tesla's autonomous cars, in Space X, or in data processing on AliExpress, and they know exactly what they're talking about, because they work with the Next-generation AI.
Elon Musk himself, along with other personalities and experts, have founded OpenAI , an initiative that aims to create AI systems that benefit Humanity, and cannot be revealed against it.
We must trust that the experts will do the right thing, and mechanisms will be put in place so that LA does not turn against us . The benefits will be endless, in fields as diverse as disease detection, cancer cure discoveries, solutions to climate change, and much more.
It is going to change the world even more profoundly than the Internet or mobile phones. Are we ready for it?
Following FAQs that were addressed thoroughly in this AI article. We hope this will add value to your learning:
What is artificial intelligence? Why is artificial intelligence important?
- What is artificial intelligence
- Artificial intelligence definition
- artificial intelligence examples
- Types of artificial intelligence
- Best definition of artificial intelligence
- Artificial intelligence applications
- Artificial intelligence future
- Artificial intelligence article