Basic Concepts of Artificial Intelligence

Introduction to Artificial Intelligence and its basic concepts

When we talk about Artificial Intelligence, it often evokes a world of robots or very futuristic technologies. However, AI is already part of our daily lives.

An example of this is when we make online purchases, where self-learning algorithms analyze our behaviors to recommend products that adapt to us; or Google Lens, which by uploading an image, identifies you and links the products that appear in it. Below is an image testing Google Lens with our office in Barcelona:

Artificial Intelligence (AI) in English, is the advanced field of current Computer Science dedicated to automating behavior commonly associated with human intelligence. And like any complex science, many concepts stem from it. Therefore, with this article, we want to start from the beginning and help you understand the basic concepts of AI.

In this article, we will talk about:

• Machine Learning
  • Pedro Domingos' 5 tribes of Machine Learning
    • Symbolists
    • Connectionists
    • Evolvers
    • Analogizers
    • Bayesians
  • Types of Machine Learning
    • Supervised Learning
    • Unsupervised Learning
    • Semi-supervised Learning
    • Reinforcement Learning
• Deep Learning
• Neural Networks
• Cognitive Computing

Deep Learning

Based on the approach of the connectionists, the branch called Deep Learning (DL) emerges. It is a subtype of Machine Learning algorithms that are based on neural networks for processing data in cascade. The term "deep" refers to the number of hidden layers in the neural networks.

Neural Networks

Neural Networks (NNs) belong to the family of Machine Learning algorithms and are inspired by the functioning of neurons in the human brain. They are based on the idea that given certain parameters, there is a way to combine them to predict a specific outcome. Data passes through different layers where a series of learning rules are applied until reaching the last layer where the results are compared with the "correct" result, and the parameters are adjusted based on the "weight" function given in each rule. Once the network has learned, it can freeze its "weights" and operate in recall or execution mode.

Cognitive Computing

And finally, some authors mention Cognitive Computing (CC) as another variant of Artificial Intelligence. These are systems that assume specific tasks or make decisions as assistants or replacing people, as they can handle ambiguity and vagueness, and have a high degree of autonomy within their area of knowledge.

Conclusions

As we have discussed, AI is already here, but there is still much to be done, not only in terms of how to exploit its full potential to bring it closer to human intelligence, but also in terms of how to control misuse.

The goal is to prevent what Elon Musk already predicted, "machines could start a war by publishing fake news, stealing email accounts, and sending fake press releases, just by manipulating data," and what we have already experienced when, through AI, sexual scenes were manipulated by replacing faces with those of well-known artists.

We must be aware of the improvements that AI brings to our lives, but without forgetting that we must use it prudently.

Deep Learning

Basado en el enfoque de los conexionistas surge la rama denominada Aprendizaje Profundo, Deep Learning (DL). Es un subtipo de algoritmos de Machine Learning que se basan en redes neuronales para un procesamiento de los datos en cascada. El término "profundo" se refiere al número de capas ocultas en las redes neuronales.

Neural Networks

Las Redes Neuronales, Neural Networks (NNs), pertenecen a la familia de algoritmos de Machine Learning y se inspiran en el funcionamiento de las neuronas del cerebro humano. Se basan en que dados unos parámetros hay una forma de combinarlos para predecir un resultado concreto. Los datos van pasando por distintas capas en las que se van aplicando una serie de reglas de aprendizaje hasta llegar a la última capa en la que los resultados se comparan con el resultado "correcto", y se van ajustando los parámetros en base a las función "peso" dada en cada regla. Una vez la red ha aprendido, puede congelar sus “pesos” y funcionar en modo recuerdo o ejecución.

Cognitive Computing

Y para terminar, algunos autores mencionan la Computación Cognitiva, Cognitive Computing (CC) en inglés, como otra variante de la Inteligencia Artificial. Son sistemas que asumen tareas o toman decisiones específicas como asistentes o sustituyendo a personas ya que pueden manejar la ambigüedad y la vaguedad, y tienen un alto grado de autonomía dentro de su área de conocimiento.

Conclusiones

Como hemos comentado, la IA ya está aquí, pero aún hay mucho por hacer, y no solo respecto a cómo explotar todo el potencial que tiene para que se acerquen cada vez más a la inteligencia humana, sino también respecto a cómo controlar el mal uso.

El objetivo es evitar que ocurra lo que Elon Musk ya predijo “las máquinas podrían comenzar una guerra publicando noticias falsas, robando cuentas de correo electrónico y enviando notas de prensa falsas, solo con manipular los datos”, y que ya hemos vivido cuando, mediante IA, se manipularon escenas sexuales reemplazando el rostro por el de artistas conocidos.

Debemos ser conscientes de las mejoras que la IA supone en nuestras vidas, pero sin olvidar que debemos utilizarla con prudencia.