That beast of artificial intelligence that responds to the name of IBM Watson is undoubtedly one of the most interesting technological projects launched by the American multinational in recent decades.

If it is little more than a curiosity capable of beating any human being in a television contest, it has proven to have enormous potential to revolutionize research fields such as the biochemist or the doctor, has helped companies make better data-based decisions and he is getting us to start “talking” with the machines naturally.

But what is your story? How has IBM Watson probably become the fastest growing IBM business division? What does it contribute to the field of AI and what do we expect from this development in the future?

A television debut

To find the origins of Watson, we have to go back to 1996, when IBM presents Deep Blue , the first AI capable on paper, of “intellectually” overcoming the capabilities of a human being and demonstrating its competence a year later by defeating the Chess world champion Gary Kasparov with a result of 3.5 by 2.5 , while in his confrontation the previous year had resulted in a clear victory of the Russian champion by four games to two.

Fifteen years later (2011) and as without a doubt many know, the IBM engineer, David Ferrucci , proposes the “most difficult yet” and leads a research team whose goal is to develop a platform sufficiently smart to beat the world champion of Jeopardy! , a popular television contest in the United States, based on the classic dynamic of questions and answers.

But why Jeopardy! precisely and no other? Not only because it was a tremendously popular contest, but because it had a devilish structure, different from the traditional question and answer program. Instead of relying on “academic” questions such as “What is the capital of Angola?”, Jeopardy! He presents to the contestants some answers, which they must know what they represent.

In other words, given the information “I cross several countries in South America and my length exceeds 7,000 km” the correct answer would be “I am the Amazon River” but in addition, the program uses all kinds of puns, language turns and unexpected combinations that make it almost impossible to find information “indexed.”

The natural consequence of the above and that explains that before the new IBM no other computer dared to face this challenge is that “the machine” that wanted to win the game was not enough to “search for information” on a computer basis.

data but somehow, he had to really understand what they were asking, capture the irony and even have a sense of humor.

The result, an IBM Watson, whose name refers to and honors Thomas J. Watson , founder and first CEO of the company.

Overcoming the challenge

In order to beat two of its most legendary champions (Brad Rutter and Ken Jennings) the Watson team builds a platform on which to apply the latest advances of the moment in fields such as natural language processing , information retrieval, knowledge representation, automated reasoning and machine learning with the objective of answering open and public domain questions.

Already in its first version, IBM explained that Watson was able to “use a hundred different techniques to analyze natural language, identify sources or find, merge, generate and classify hypotheses.”

which earned him the prize of one million euros by finally imposing a slight difference on his two human opponents.

After demonstrating its capacity on the pitch, in 2013 IBM developed its first commercial project together with the Memorial Sloan Kettering Cancer Center hospital in New York: a practical application that greatly improved decision making when managing clinical treatment (and pharmacological) of patients affected by lung cancer.

Since then, Watson’s capabilities have not stopped growing and over time it has incorporated new technologies such as the power of the cloud to go further and has ceased to be a computer system of questions and answers to a very different one capable of «seeing »,« Hear »,« read »,« like »,« interpret »,« learn »or« recommend ».

Elementary, dear Watson

Of course, the IBM Watson that David Ferrucci’s team presents in 2011 is not exactly the same one that companies and researchers from all over the planet are working with today.

At this time, keeping this computing giant in operation has led to the start-up of a cluster consisting of 750 IBM Power 750 servers , each of which uses an eight-core Power7 processor at 3.5 GHz. to add no less than 16 TB of RAM.

This means a gross computing power of 80 petaflops , which in terms of supercomputers (in many ways does not fall into this category) would enter directly into the ranking of the top five .

How does all this raw power move? If we examine its software layers we discover that IBM Watson is based on the DeepQA artificial intelligence platform developed by IBM and based on programming languages ​​such as Java, C ++ and Prolog.

To support it, it uses the Apache UIMA (Unstructured Information Management Architecture), SUSE Linux Enterprise Server 11 as operating system and Apache Hadoop framework for distributed computing tasks.

With all these ingredients, the IBM guys get Watson to be able to process 500 GB (the equivalent of one million books) per second.

In fact, as sources of information input, the platform draws on all types of encyclopedias, dictionaries, scientific papers, taxonomies, ontologies … with projects such as DBpedia (a project to extract structured information from Wikipedia), Wordnet or Yago .

Watson enters the big door of business

The collaboration with Memorial Sloan Kettering Cancer Center was the first of many that came next. The intention of developing Watson’s business capabilities was reflected in 2014, with the launch of the “IBM Watson Group” a new business division headed by Michael Rodin and which would begin to serve more than 2,000 people.

Since its launch, IBM has invested more than $ 1 billion in the division, which has focused its efforts on the development of three major growth branches: Watson Discovery Advisor, Watson Engagement Advisor, and Watson Explorer.

The first, Watson Discovery Advisor works on research projects related to the pharmaceutical and biotechnology industry.

Watson Engagement Advisor is the area from which self-service applications (chatbots, virtual assistants, etc.) are developed, based on the interpretation of natural language, responding to the needs of business users.

Finally, Watson Explorer helps users discover, structure and share information from huge volumes of data.

And what projects have you been involved in lately? Without going any further, in the last US Tennis Open, Watson was the platform on which the intelligent assistant «Ask the US Open» was built , which thanks to his conversational abilities informed fans who attended the tournament about everything they had to know to improve their experience: “before going” tips, points of interest in the stadium, integration of maps, results and live scores, weather alerts, etc.

In the financial field, the Italian bank Creval has achieved that IBM Watson reduces up to 80% the requests for information that its call center receives and that need to be answered by a human being and by collaborating Thomson Reuters is getting companies to comply without problems with all regulations that have been put in place to manage the information and privacy of all types of customers, as in the effective application of the GDPR).

And not only. In the experimental field, he has developed, for example, an “electronic language” (IBM Hypertaste) that, inspired by the functioning of the sense of human taste, has the ability to detect and analyze different types of liquid quickly and without having to go to a laboratory, or medical projects as interesting as IBM Watson for Oncology, able to combine the experience of oncologists in cancer care to help doctors consider individualized treatments for their patients.

Watson in Spain

How could it be otherwise, the technology that IBM provides with Watson has aroused the interest of not a few Spanish companies. We highlight below some of the most interesting:

Cerealto Siro Foods , which launches more than 250 new cereal products on the international market every year, has incorporated IBM Watson’s artificial intelligence into its innovation process to detect early trends and as a source of inspiration and has already launched its first snack « based on artificial intelligence »: rice pancakes with peas and lentils.

Endesa has created a virtual assistant using IBM Watson Assistant technology that allows customers to get answers to their questions immediately, making it easier for center agents to spend more time on more complex requests and selling products or services from the power company .

SEAT has presented “Mobility Advisor” , which uses artificial intelligence from IBM Watson to help urbanites to make informed decisions about different transport options during their day-to-day: from cars, motorcycles or bikesharing to public transport, so that the option chosen is the fastest and most efficient.

Helvetia Seguros has decided to facilitate the purchase of family assistance insurance with the help of a virtual assistant.

Clients talk in natural language with the virtual assistant accessible on the Helvetia website, who collects the relevant data and offers a customized quote, reducing the time, days, of a possible traditional contract to minutes.

Not to mention the Tax Agency itself , which since July 2018 helps companies to meet their tax obligations by allowing them to rely on IBM’s artificial intelligence to comply with the Immediate Supply of Information (SII) of VAT, which currently requires the electronic sending of the record of your invoices during the 4 days following its issuance or accounting

But there are plenty of examples. The Prisa Group collaborates with IBM to offer better meteorological information to its users, Carolina Herrera reinforces thanks to this technology its e-commerce strategy and the secondary education institutes of the Community of Madrid and employ Watson to help students train in programming and development of AI applications.