Nowadays, Artificial Intelligence is an enigmatic label that in many cases is generalized, alluding to any capacity by which machines can perform tasks that were considered solely human.

It is difficult to understand this concept if we do not think about specific examples, and when we go deeper into the different aspects that AI can cover, we realize that it is a very polyhedral phrase which, among others, encompasses phenomena such as:

• recognition, transcription and reproduction of the human voices and sounds,
• language processing and its generation,
• artificial vision and
• automated image recognition,

To all of those, we can add concepts that dot the articles and reports on the latest trends, such as "machine learning" or "automated learning", and "Deep learning" or deep learning based on the emulation of brain systems with algorithms based on neuronal networks.

All of this mixed with data in multiple forms, such as big data, data analytics, data visualization, data science, makes it difficult, in many cases, to glimpse their use cases and possible applications in the real world.

Several factors come into play that are pushing companies to invest, develop and transform their processes using artificial intelligence:

Technology Itself

The time for AI is now! And this is mainly due to the combination of 3 different factors:

• data amount and availability on the cloud;
• software and algorithms that are open, mature and available for everybody;
• and hardware, as GPUs have been invented and has evolved very quickly, allowing to create very powerful machines to process a big amount of data in a very quick way

The Amount of Massive Data Generated...

...of which it is calculated that a percentage of between 80-90% is unstructured. In 1992, daily global internet traffic was 100GB / day and in 2015 it has grown to 15k million GB per day.

By 2020, it is expected to reach about 44 zetabytes of daily data generated, and yet most of the data that is produced is not analyzed, and the unstructured, mostly is not processed.

An example of unstructured data is text. Natural Language Processing technologies could, in many cases, be used to transform these unstructured linguistic data into knowledge and obtain added value "insights".

This was due to the classification, extraction and understanding of the information, that the human mind cannot get, especially when the volume of information is high. .

IoT

One important point for the acceleration of this new digital revolution is the proliferation of digital artifacts, or the so-called "new IoT (Internet of Things) platforms".

This which allow users to constantly interact not only with their smartphones or computers, but also using interfaces that in many cases could go beyond traditional screens, as wearables or pieces that are added to cars or systems, which are activated by voice, sensors or different kind of external inputs.

All these gadgets help us generate data, classify them and use them to build the models which AI uses for feeding the algorithms.

All these circumstances have made us face a true revolution with great consequences, of which we are only seeing the tip of the iceberg, but which are not only aimed at transforming and improving current business models.

This is achieved through cost reduction and automation of processes but to generate new models and new lines of income thanks to the breadth of possibilities, the generation of new insights and the improvements in data analysis both in volume and speed in real-time.

It is expected that the benefits of an investment in AI will be high in the income accounts of different businesses in different sectors:

• The average cost savings that have been observed in companies that apply this type of artificial intelligence solutions is about 20%, and in some cases, much more.

• The use of AI generates operational efficiencies, which allows human beings to be relieved from automated tasks that can be done by a machine, reducing the costs of processes and giving humans the possibility of using their knowledge for more specialized purposes.

• AI is not only for cost-cutting, as it allows the generation of new knowledge thanks to the insights that can be obtained by analyzing thousands of data by using Machine Learning models on them.

• The third step in the transformation is the creation of new business models, based on the data generated in the previous steps.

The intelligent processing capacity of thousands of data in real-time poses new challenges. AI will change business approaches in a very short time, and across a wide variety of industries.

However, although technology is already mature and is working very well for specific purposes, its application across the different industries is uneven.

But Artificial Intelligence is not “one technology” by itself, and it does not work in the same way for all the different cases and situations. It is a combination of different technologies that vary, depending on the type of problems that are to be solved and on the type of data that are behind those problems. For example:

• Language data are processed as text and/or voice, but both technologies are different.

• The data below the text processing are text strings, however, the voice assistants need to convert first the sound waves into text for analyzing them.

• Natural Language Processing uses rules, dictionaries and grammars to make computers “understand” the sentences and word structures; image processing uses pixels as the basis of the analysis for its algorithms.

One common point of all these technologies is that they “learn” and improve their models, and the learning process (Machine Learning) might be done with human inputs (supervised learning) or by themselves (unsupervised learning).

The application of neural networks as the algorithmic solution for unsupervised learning is called “Deep Learning”. It is a very powerful technology that has proven excellent results, but only in specific environments with a lot of data to train the models (for example medical image recognition or image classification).

Understanding AI means knowing use cases where it can be applied. Let’s focus on one of these technologies, which are getting famous in recent years: language technologies that are on the basis of chatbots and virtual assistants such as Alexa or Cortana.

Where can these cognitive engines or artificial intelligence platforms really be applied to our markets, to different industries, and specifically, to companies that work with linguistic and textual data?

A.I. systems are capable of:

1. Extracting the most relevant information from a text.
   
2. Making a brief summary.
   
3. Streamlining a FAQ system within a large database.
   
4. Interacting in the form of a chat with questions and answers about a textual corpus.
   
5. Detecting and analyzing the feeling of a customer when reaching out to a call center.
   
6. Active listening in a multichannel system in which social networks and press publications intersect with those of industries.

To give some specific examples of existing use cases, we find that there are:

• Digital lawyers who search, analyze and process sentences by analyzing documentation in an automated way without reading them, but extracting the most relevant points and summarizing them.
  
• Automatic systems for classification and analysis of medical records in the health sector.
  
• Systems for the automatic generation of reports from the data provided by the Smart-grid in the energy sector.
  
• Robo advisors that already interact in natural language with humans for financial recommendations.
  
• Virtual assistants that allow streamlining the policy consultation process in insurance companies.

So the question remains, who will be able to resist the challenge of artificial intelligence? And who would fall behind?