The incursion of artificial intelligence into society is not a new phenomenon. On the other hand, the ChatGPT revolution, which became the application with the fastest growing number of users in history two months after its launch, yes.
The bombardment of news since the appearance of this text-generating bot in November 2022 has spread information about terms that not everyone is familiar with. So that the novelties in the field of artificial intelligence are not conceived as a foreign world, we define below some concepts related to this field of computing.
Artificial intelligence
Field of computing focused on the creation of machines capable of performing tasks that require intelligence, such as learning, decision making or pattern recognition. An intelligent machine would be one that, with the use of algorithms, has been trained to be able to receive information, process it, learn from it and respond accordingly autonomously.
The term is attributed to the computer scientist John McCarthy, who coined it in 1955 around what is known as the Dartmouth Conference. The event brought together the field of computer science and other related areas with the objective of discussing the possibility of machines that behaved in this way.
Algorithm
An algorithm is a set of precise, defined, and ordered instructions -a logical sequence of instructions- that are used to solve a problem or carry out a specific task. It would be like the assembly manual for an Ikea piece of furniture that, after following a few steps in a certain order, would make it possible to get a new table, for example. Unlike these instructions, the algorithms are not written in natural language, but in different programming languages, such as Python or Java.
Machine
A mechanical, electrical, or electronic device used to perform a specific task or to perform a series of tasks automatically. Machines range from the simplest, such as a lever or pulley, to the most complex, such as a computer. The sophistication of machines over time allows us to talk about intelligent machines, among which are autonomous vehicles.
Bot and chatbot
A bot, a term from “robot”, is a computer program that performs repetitive, predefined and automated tasks in digital environments. Bots are used for a wide variety of functions ranging from replying to emails to web crawling to chatting with users.
Chatbots or conversational bots belong to this last category. These types of bots have been designed to interact with people in a natural and conversational way thanks to natural language processing and other artificial intelligence techniques. Chatbots can be programmed to understand and respond to a wide variety of questions and requests, as well as improve their responses over time. Given their characteristics, they are often used to provide information automatically, as occurs in customer service. ChatGPT is a chatbot.
ChatGPT y GPT
ChatGPT is the new and revolutionary chat system developed by the artificial intelligence research company OpenAI. This application has the ability to process and analyze information in order to respond to user requests. To do this, it uses the GPT (generative pre-trained transformer) language model, a natural language processing model trained with large amounts of text data.
GPT is based on a type of architecture -model structure- called Transformer and is used for a wide variety of tasks such as language translation, text generation or reading them. GPT already has several versions, GPT-4 being the most recent to date, released in March of this year.
Machine learning or machine learning
It is a branch of artificial intelligence focused on giving computers the ability to learn from data and identify patterns, without being explicitly programmed for it. In this way, computer systems can make predictions autonomously and improve their performance over time.
This technology is applied in a wide variety of fields, from robotics to medicine. Virtual assistants like Siri, Alexa or Google Assistant use machine learning to recognize speech and provide accurate answers to complex questions. It is also used in recommendation systems on platforms such as Netflix or Spotify to analyze user consumption patterns and offer personalized recommendations.
Deep learning or deep learning
It is a type of machine learning that focuses on imitating the learning of people. Deep learning is based on artificial neural networks, which are mathematical models inspired by the human brain used to process large amounts of data and extract complex patterns.
Artificial neural networks
Artificial neural networks are mathematical models inspired by the functioning and structure of the human brain. These networks are made up of a set of interconnected nodes or neurons that are organized in hierarchical layers in order to process information in a similar way to neurons in the human brain, which implies a high level of abstraction and complexity. In the same way as in the organ, each node receives inputs from other nodes and produces an output that is transmitted to other nodes in the network. The connections are weighted, which means that the strength of the connection is adjusted based on the training data.
natural language processing
Branch of artificial intelligence focused on providing machines with the ability to understand and generate human natural language, which makes it possible to improve communication between people and devices. In this way, computers can perform tasks such as showing the structure and meaning of texts or interacting conversationally. An example of use, beyond ChatGPT, is found in Google Translate, which uses natural language processing to automatically translate text from one language to another.
Fourth Industrial Revolution or Industry 4.0
The term industry 4.0 refers to the digital transformation of industry and manufacturing marked by technologies such as the Internet of Things, artificial intelligence or robotics. The founder and director of the World Economic Forum, Klaus Schwab, wanted to go further when talking about a Fourth Industrial Revolution, to indicate the fusion of new technologies and systems and their interaction through physical, digital and biological domains.
The terms have been widely used ever since to describe the convergence of digital and physical technologies that are changing the way businesses, the economy, and society in general work. One of the trends of this new revolution is the automation of manufacturing.
Internet of things (Internet of Things, IoT)
It refers to the network of physical objects connected through the Internet, which allows them to communicate and collect data for further analysis without the need for human interaction. An example of this would be a smart home, where electronic devices such as lights, thermostats, appliances or security cameras would be connected in this way.
The collection and analysis of data in real time can generate useful information to make decisions in different areas, such as, returning to the example of the smart home, adjusting the temperature automatically based on the patterns of use of the house and the climatic conditions.
Big Data
Analysis and management of large volumes of data that are very complex and difficult to process with traditional data management tools. The term also includes the characteristic of said data, such as the speed at which it is generated, and hence its great volume, or its complexity and diversity. Companies and organizations can use the processing and extraction of information from this data to make more informed decisions, as occurs with the information generated in social networks such as Facebook, Instagram or Twitter. Such activity is called data mining or data mining.
Turing test
The Turing test is a test proposed by the British mathematician Alan Turing in 1950 to determine if a machine can be considered intelligent. The test is based on an imitation game, in which a human interacts with two participants: a person and a machine. To be considered intelligent, the machine would have to imitate the person well enough to fool the judge into thinking that it is, in turn, a person.
Prompt
In the context of artificial intelligence, a prompt refers to a text input or instruction that is provided to a language model in order to generate a specific response or action. It would be the stimulus for the artificial intelligence to respond, like a question or a description to generate an image.
Hallucination
In artificial intelligence, a hallucination refers to an error in a machine learning model that generates false but highly detailed and convincing information. This can happen by biasing the model in a way that makes the wrong connections between data and patterns. In other words, it would be a poorly established pattern that leads to invented answers. Hallucinations are therefore different from random errors, since they are repeated and persistent phenomena that indicate a deeper problem in the model.
generative artificial intelligence
Branch of artificial intelligence focused on the generation of original and new content in the form of images, music or text, among others. These systems use deep machine learning models to learn how to create consistent and relevant content based on what the user requests.
Generative artificial intelligence is based on the use of two neural networks: a generator and a discriminator. The generator generates the new content from the input data that is similar to the training data. The discriminator reviews the creations and compares them to the training data to determine if they fit what is being searched for.
Software
Set of programs or instructions that allow an electronic device to perform specific tasks. A software is developed by writing such instructions in a specific formal or programming language. These are compiled into a file that can be executed by the physical part of the computer, or hardware, to perform the task at hand.
Byte
A byte is a unit of measurement of information used in computing and telecommunications. It represents an ordered set of 8 bits, these being the minimum unit of information, and is equivalent to a letter or another alphanumeric character. Storage in most computer systems or computers is counted in bytes. Therefore, the size of files or programs that it measures in this way.
Expert systems
These are computer programs or systems capable of emulating the reasoning and decision-making of a human expert in a given field. In this way they can solve complex problems that require the expertise of such a professional. Such systems have a database from which they make decisions regarding the problem posed thanks to a set of inference rules. An example is found in medical diagnostic software that, based on the knowledge they have on the subject, serves to help professionals make an accurate diagnosis.
Deepfake
A deepfake is an artificial intelligence-generated video, image, or audio that imitates a person. The resulting content is therefore fake, but so realistic in appearance that it can easily pass for real. A couple of recent examples of deepfakes are the image of Pope Francis dressed in a padded white Balenciaga coat or the images of the alleged arrest of former US President Donald Trump. Both were generated with the Midjourney artificial intelligence program. As a consequence, those responsible for the software made the decision to end the free trials.
Phishing
Cyberattack technique generally consisting of sending fraudulent emails used to obtain confidential and personal information. The emails usually ask the victim to enter data such as their credit card number, their ID or the password to access online banking, or to click on a link to an apparently authentic malicious website.
black box
It refers to an artificial intelligence model or system that performs a specific task, but without fully understanding how it did it. The internal processes used and the multiple factors remain unknown. Not even programmers or administrators of the machine or the algorithm know how the result was arrived at. That is, you can only know the input and output of the model and not what happens in between, which remains a black box.
A similar phenomenon occurs in the aforementioned neural networks. Although it is known that the neural network receives an input and produces an output, the intermediate process is difficult to understand due to the complexity of the model.
