In recent posts, we have spent a lot of time dealing with machine learning and cognitive systems. The name IBM Watson has also been mentioned several times. But what is IBM Watson? And even more important: What can it actually do?
Data-driven cognitive systems are different from the programmable systems we have been using for more than 60 years. Almost all computers in use today are based on the Von Neumann architecture defined by mathematician John von Neumann in 1945.
The Von Neumann computer processes the externally entered programs and data sequentially, with processing taking place in the order of the stored program instructions. Although this has helped the government, entrepreneurs, scientists, consumers, etc., this principle is very limited and not very expandable.
And especially in today’s world, we need systems that can process real-life data quickly and that don’t require laborious programming – because they should function independently. As independently as the human brain. The only difference, however, is that while the human brain needs 20 watts to think and work, Watson needs 85,000 watts. 85,000 watts are needed to evaluate all the complex data that is on the Internet and make it usable.
Just as we invented industrial machines to augment our strengths and work faster, we now need to develop a new generation of machines to augment our cognitive capabilities. And indeed, the architectures of such cognitive systems have more in common with the structure of the human brain than with those of Von Neumann’s classic machines.
The intelligent handling of large amounts of text is Watson’s strength. For the famous episode of the quiz show Jeopardy! In February 2011, in which the system beat two human opponents, Watson had to absorb a database of 13 terabytes.
Statistical data analyzes are not carried out, as is the case with conventional analytics tools. Statistical methods are used internally by Watson to be able to work at all, but the system builds neural networks and large index structures to create correlations. Patterns are extracted and stored in their own network structures – Watson does not carry out statistical analyzes of the data.
Why? Because in this case the user is not using an application, but rather receives a natural answer to a natural question. This leads to an unprecedented, active collaboration between man and machine. Watson evaluates the data, abstracts it and provides the answer in the user’s language. Compared to conventional analysis tools, this is a revolutionary approach: Watson combines machine learning and other artificial intelligence technologies in one system that can be accessed via various apps.
And the best thing about Watson is that it is very versatile and can be used in many different ways. Whether in the financial sector, healthcare, customer service or everyday life – Watson can do it all.
Because no matter what field you are in, the amount of information available is enormous. You have neither the opportunity nor the time to fight your way through the data jungle.
In the financial sector alone, there are 9,000 pages of financial news every day from the Reuters news agency. Added to this are market research reports and documents from several companies and from Wall Street. How can one person be able to comb through all this data day after day?
Watson is very easy to use. It is similar to Apple’s voice recognition software Siri or the voice recognition on Android devices. You speak to the system – and the system responds.
For example, there are over three hundred different types of cancer, from leukemia to colon cancer. Each type of cancer has its own therapy and treatment options and changes with the patient.
A doctor who doesn’t have time to read through hundreds of books and other sources of information for a cancer patient could ask Watson: „What are the treatment options for colon cancer?“ And Watson would – after evaluating the patient’s data such as age, previous illnesses, family history of the disease, allergies, etc. – find a suitable, individual treatment option for this patient. In seconds. This allows the doctor to act more quickly but also ensures that the doctor has not overlooked or not included any method.
And the artificial intelligence in Watson can not only understand and answer questions, but also see. IBM has taught its computer system to recognize the content of images.
We can be excited to see how Watson will be used in the future and how IBM will manage to further expand the system. One thing is certain: Watson will make people’s work easier in every field in the future and will even be able to help us with the smallest of things in our everyday lives, such as cooking.
Cognitive systems like Watson have the potential to change the way companies think, act and work. These systems learn through interactions and provide evidence-based answers that drive better outcomes. (IBM on Watson)