The Super AI

The tendency to generate greater productivity with less effort seems to be inherent in the human being. This has led people to look in science how to create tools or machines that facilitate these processes, such as in the first industrial revolution or the use of electricity that changed the history of the world. Today is the turn of Artificial Intelligence (AI) technology, the 4th Industrial Revolution, which is a strategy between several branches: from physiology, philosophy, ethics, mathematics, statistics, and applied engineering.

Thanks to science fiction, there are many expectations, and AI has taken several concepts that are becoming a reality every day. But before giving a talk about what AI could do, we must define the following: What is AI? What are its expectations? Why is it considered an industrial revolution? Why is it important for a company to use AI to transform itself into an AI company? And what about AI business ethics and responsibility? With these doubts, I want to explain to those involved in AI what it is made of, how society’s lifestyle will change, and give a start on how to be involved in this new industrial revolution. For this first blog post, I will try to give an induction of AI and its basic concepts and the influence it is currently presenting.

What is cloud computing? Find out the main benefits of cloud computing and how they impact the telecommunication industry.

What is AI?

A basic definition of AI is the ability of a computer system or software to analyze different types of data and use this for decision-making, and that is based on prior learning acquired in an automated way. This is not the only concept that exists, each day that advances, new knowledge, and specializations are acquired, shaping what we know today as AI.

AI has been promoted thanks to autonomous learning. The most used is supervised autonomous learning, where a system is assigned an input A and its output B, such as detecting a person in a photo, identifying if an email is junk or spam, the translation of the text or recommendation system, at an industrial level to detect whether a product is a factory defect, etc. The boom is just beginning thanks to 3 factors: level of computational processing, improved software tools and algorithms, and access to large amounts of data that is most important. The internet and large data storage systems currently allow an improvement in the results, although these must be intelligently selected, which still involves a human factor. Now, the data can come from images, sensors, email, and the vast majority must be processed and vectorized. This would be input A and output B should be the result of the input. In many cases, the result B must be put manually in the autonomous learning training.

With large amounts of information, the autonomous learning algorithms you are looking for are relevant data that help make statistical predictions in a subsequent event with only one among A. Another important piece of data is that AI and autonomous learning are statistical processes, that is, they make percentage predictions, which means that there are no absolute values ​​as an answer. Although they help, not only does having huge amounts of data guarantee success, but it must also be a strategy for implementing AI.

The Basics

We are often told about AI solving many problems. Still, within it there are several terms such as “autonomous learning, data science, neural networks, and deep learning” and they are general terms that together make up an environment called AI and everything is essential.

To easily define these terms I am going to present the following example: Suppose I have information on automobile flows at the exit of a city, such as the number of automobiles on the specific day and time, the type of automobile, if it is a medium-sized car, bus or truck, the weather, amount of population, stratum if it is a weekend with a holiday, etc. Autonomous learning generates a system that inputs A to the vehicle data and can generate an output B that can be the traffic density, which is very useful for organizing an operation to improve traffic on the way out of the city. Autonomous learning before being able to give a result must be trained with data, these data are not chosen at random and there must be an analysis to identify more relevant data, a data science team carries out this process. They are in charge of providing processes and filtering the most important data into the neural network.

A neural network has nothing to do with how the brain works but they were an inspiration. Today a neural network is a great mathematical equation that with input A calculates the vehicle density B at a certain time of day. The term neural network has changed and evolved into deep learning, which is different methods and, depending on the problem to be solved, it changes; “Supervised and unsupervised learning, reinforcement learning, graphic models, planning, knowledge, graphs, among others”. All this helps to generate the mathematical equations of a neural network. The following graph can help us to summarize better.

If you did not know today, there is a large AI working in different services, internet advertising, search engines and translators, virtual assistants like google now, Siri, Alex, etc. All these systems are designed to solve specific problems and although we want to be optimistic, they still have limitations.

On the other hand, the change is so fast that as it grows, it takes more concepts and each one specializes and expands. I hope that with this induction you can be clearer about what AI and its components consist of. For the next blog, I will explain how it impacts society and how to transform a company into AI.

What is cloud computing? Find out the main benefits of cloud computing and how they impact the telecommunication industry.