How does IBM Watson work?

Watson is a supercomputer developed by IBM to combine advanced analytical software and artificial intelligence, which together provide a smart service to answer questions. It is named after IBM founder Thomas J. Watson. Of course, we've seen a glimpse of such a machine in our smartphones, car GPS or other voice-activated devices, but when it comes to answering questions, Watson is the undefeated king.

Watson's architecture

Important parts of Watson include:

• UIMA (Unstructured Information Management Architecture) is a software library developed by IBM and the Apache Software Foundation
• Hadoop, an open source framework for processing very large volumes of data in parallel on many computers, developed by the Apache Software Foundation.
• SUSE Enterprise Linux Server 11. It is currently the fastest Power7 processor operating system.
• 2880 processor cores.
• 15 terabytes of RAM.
• 500 gigabytes of knowledge bank.
• DeepQA software developed by IBM. The software is designed to combine machine learning techniques with language processing software to create a search engine for answers to questions.

With Watson's hardware, it is capable of computing at an impressive rate of 80 teraflops (trillions of floating-point operations per second). To use all that capacity in practice, Watson is given access to 200 million pages of information from a cluster of 90 servers. All this information compares it to 6 million logical rules.

For all of that, however, it requires a certain amount of space (Enough for about 10 refrigerators) for storage. All this to create a machine that surpasses man's ability to answer questions; Watson is logical, factual, able to retrieve data in a split second, and certainly faster than humans to answer questions.

How Watson finds answers

Like a human brain, Watson requires huge amounts of data to shape its response. All data (or related materials) is entered into its knowledge bank. The data can be web pages, documents or PDF files. To understand issues, Watson leverages the power of its language processing software to help understand grammar and context. Each sentence or question can have multiple meanings, but Watson tries to figure out the actual purpose of the question after a thorough language evaluation.

Data to generate answers is mostly unstructured, and so Watson must initially do an analysis to extract a lot of possible conclusions. In finding an answer to a question, it ranks all the conclusions based on giving each of them points for the relation to the question as well as the validity of the supporting evidence. As Watson is fed new data, its knowledge base is updated automatically.

Jeopardy

In January 2011, IBM decided to showcase Watson's ability in Jeopardy! In competition against two human contestants: Ken Jennings and Brad Rutter. Ken had the record for the longest series of winning shows with 74, while Brad was the contestant who had earned the most ($3.25 million total) from the show. Watson won.

How you can use it in your business

Watson has been around for a while and various companies have begun to leverage the technology. Existing customers include many business areas: construction, hospitals, and popular fashion apparel manufacturers. Some universities also use Watson for research purposes.

In some respects, Watson is like any typical search for e.g. Google, but specialized to provide answers to questions. Instead of providing countless options in the form of search results for a search, Watson provides only the best-substantiated answer to a question.

Watson also understands that not all data is created equal, so it analyzes them to narrow the possibilities. In the same way that a person would not rely equally on all sources of an answer, Watson will rely more on certain sources than on others.

IBM has designed Watson based on a "cognitive intelligence" approach and has spared neither the development of the advanced software nor the powerful hardware that Watson runs on. The result is a computer that processes information in a way very similar to the way a human brain would do it. It has the great advantage that we as humans can also relate more easily to how it works, even without technical insight into the details.

Watson has been used in a number of companies for various purposes:

• Mears Group: A UK provider of care and care services. Since 2015, the company has been integrating with Watson to get suggestions for opportunities for better revenue and faster reporting. Mears Group has around 20,000 employees and more than 700,000 social housing benefits across the UK.
• Citigroup: Banking company. Used Watson's analytical capabilities in 2012 to get suggestions for improving customer relationships. Also used it to analyze their data to find risks and opportunities in their business.
• Hartree Centre: A UK IT company, Democrata, uses the powerful algorithms at the Hartree Center research center. Democrata aims to analyze potential environmental impacts and predict the risks of new major construction projects such as highways and rail projects. Without the intelligent processing of the large amounts of data, it would be a very time-consuming manual task.

Industries at risk

One of the biggest concerns about Watson is that in the long run it can replace the need for labor in many well-known occupations. Businesses will thrive, but professionals may be forced to find other jobs. A report from the MIT Technology Review predicted in July 2014 that Watson could eventually replace human doctors in oncology. Watson could keep up-to-date and current on data and research progress to always provide the best treatment - something that, with the explosion of knowledge we have seen, has become very difficult for human doctors.

Watson may be considered "disruptive" technology for many established companies, but it is always best to see it as a huge opportunity to grow your business and provide better customer satisfaction.

Indeed, the ability to make conclusions based on unstructured data can help (or replace) jobs in industries such as travel agents, investment advisors, bank advisors and real estate agents. And if Watson or other similar systems based on artificial intelligence really replaces the need for these jobs, one will have to change the human role in business; even the smartest employees won't be able to compete with Watson's speed in information processing.

Fortunately, there are always the more creative elements in a profession where even the smartest machines can fall short. One can imagine that one's job is to create the creative combinations for one project and every day at four o'clock a system like Watson will take over to use these inputs to complete the work.