Elementary My Dear ......

Elementary My Dear ......

“Elementary, my dear Watson” or so would the greatest detective in English literature supposedly state with smug satisfaction and then proceed to walk his assistant Dr John Watson through a complex maze of facts and observations and his deductions based thereon to unravel what had seemed an unfathomable mystery to readers till that point of time. Dr Watson has been portrayed as ‘astute’ but nowhere as intelligent as Holmes. No doubt he served an important part in the Arthur Conan Doyle stories; duly recording the proceedings of each case, but I rather felt sorry for the old chap who must have felt thoroughly humiliated a number of times by lack of his analytical abilities.

Well, the fictional Dr Watson seems to have finally had his revenge in real life through a namesake of his, Watson the computer developed by IBM (named after its first CEO Thomas J. Watson).  The objective of IBM in developing Watson has been "....... to have computers start to interact in natural human terms across a range of applications and processes, understanding the questions that humans ask and providing answers that humans can understand and justify." 

Such a computer is a natural evolution of machine intelligence which IBM has been diligently developing starting from IBM 704 which was built to play checkers in the 1950’s and had an inbuilt capacity for self-learning.  In the years that followed, IBM developed Deep Blue computer which defeated Gary Kasparov in the historic match between the two in 1997 which defined the moment when human intelligence was surpassed by a computer. But the Deep Blue was a still a machine, a hugely capable machine able to compute 200 million positions per second, but still just a machine. It could follow the algorithms laid down by its programmers and evaluate the various lines of play in terms of gains and losses such moves would result in. IBM itself never claimed that Deep Blue was endowed with “artificial intelligence” though Kasparov himself did mention that he sensed something akin to “human intelligence” in some of its moves.  Deep Blue was placed somewhere halfway in the Top 500 Supercomputers list at that point of time.    

IBM Watson goes far beyond that. It was developed from IBM’s DeepQA technology which was designed for playing Jeopardy, a quiz program on NBC. DeepQA had to confront the unstructured way we humans interact with each other;  half-formed sentences, flexible syntax, colloquial terms, multiple meanings of the same word etc and, most importantly,  speech recognition and interpretation of  pictorial data . DeepQA team had to build all these functionalities into Watson’s working and more to enable Watson to continually learn from its experience, the technologies that IBM built to apply advanced natural language processing, information retrieval, knowledge representation,  automated reasoning, and machine learning technologies Watson accomplished the assignment for which it had been created without a demur and ultimately won against champion Jeopardy players  Ken Jennings and Brad Sutter who had had long unbeaten runs previously. One can safely say that the era of cognitive computing has arrived.

The processing capability that Watson has are extremely impressive to say the least. It works on a 3000 core assembly of processors with RAM of 16 terabytes and processes 500 giga bytes of information per second equivalent to about a million books per second. As an aside, high as this processing capacity seems, we humans should take satisfaction that it takes all that computing power to match the one and quarter kg of white and gray matter that each of us have been endowed with. Watson requires  as much space as eight refrigerators to do what we do very routinely as we walk and run doing our daily chores. Watson may not be able to match what we may know intuitively but as far as crunching raw data and making a dispassionate analysis without any pre-conceived notions or taboos that we hold, Watson will be the winner.  

So what does IBM propose to use Watson for? Given its abilities, Watson is best suited for crunching a huge amount of data like any of the Supercomputers would, but its ability to work with unstructured data and arrive at answers to various questions in a human-like manner is what it would give an edge as an ideal decision support system.  Such a platform which can search huge and disparate databases and extract necessary information can far surpass any human’s capabilities especially in today’s world where our knowledge base is growing exponentially.  With this sort of data crunching capabilities, Watson is sure to surprise us with finding hidden and even hitherto unsuspected correlations between two or more parameters. 

When posed a hypothesis, Watson comes up with a few alternatives indicating possible solutions with varying degrees of confidence and it also lays down the reasoning it has used to arrive at the solutions. The same process can be refined though interaction with the researcher who is seeking a solution. The whole process acts as a learning experience for Watson who can then further improve his reasoning or modify the weightages that it may have assigned to various factors. 

IBM is hopeful of leveraging just these abilities of Watson. IBM has been working with medical schools like Memorial Sloane Kettering Cancer Center to develop a system to suggest best line/s of treatment for cancer.  Ginni Rometty, Chairperson, President and CEO of IBM, explains IBM’s decision to go after cancer as a philosophy of IBM to go after the most challenging of problems that they could find.  The system demo with a hypothetical patient which you can find on YouTube will give you a good idea about the interactive process which a doctor can follow with Watson’s assistance to arrive at the most plausible line of treatment and further tests that will need to be done to improve the chances of making the treatment work. https://www.youtube.com/watch?v=hbqDknMc_Bo. 

In India, Manipal Hospitals have signed up to provide help to cancer patients through Watson. As time goes by and the word spreads about availability of Watson’s expertise, one feels that a second opinion from Watson will be a must for every cancer patient who wants to get his / her diagnosis confirmed and plan for the best treatment under the circumstances specifying the constraints that each patient is likely to have. I am sure that over next few years any cancer hospital worth its name will sign up for this service.    

IBM has identified a number of fields where the Watson platform can be employed to improve the decision-making. Genetics with its huge data is one such field. True we have mapped the human genome, but just consider this. Each human being is different from any other human as far as actual genetic code is concerned, to the extent that you and me, each have uniquely different variations in our genes to make each one of us the only person to have ever walked on this earth with this particular genetic  map. We therefore need to do mapping of a lot more individuals to begin to understand the variations that are possible in different ethnic groups. Then there is a question of how each gene functions or “expresses” itself, and the sequential interactions that it has to have with numerous other genes to express itself.  A huge lot of work is being done in this field by a veritable army of scientists spread all over the globe. It is doubtful if any human can hope to keep track of the torrent of data that these developments are throwing up.  With Watson’s data analysis capabilities, it would be possible to do that and bring the goal of personalizing medical treatment for individuals a bit closer

.

IBM is encouraging a wider acceptance of Watson platform by inviting industry partners to work together and develop apps tailored to their own needs. A number of companies are already actively using Watson for applications as diverse as hospitality, music, travel, financial services, call centres and even analysis of satellite imagery for resource management.  A novel example of leveraging Watson is that of Prof Ashok Goel, of Computer Science department at Georgia Tech who used a bot developed on Watson as a Teaching  Assistant. The students  interacted with the bot through the term without knowing that Jill Watson they were dealing with was an artificial entity. Jill Watson, by the way, was named by the students as the best among the 10 TAs who were assisting Prof Goel for that term.

Major usage of Watson would be in fields where inputs to decision making are extremely unstructured for example legal, governmental regulations. It will indeed be a great help to the judiciary if Watson can analyze whole range of legal records (both national and international) and point the judges in the right direction rather than depend upon cases cited by litigants. 

As long as data can be gathered within acceptable range of accuracy, Watson can be brought to bear upon tackling developmental problems.  In fact IBM has already embarked upon a program supported by Watson to assist African countries in healthcare and education sectors. 

As stated earlier apart from the governmental sector, large private organizations will be greatly helped by Watson. Top management of global conglomerates working across industries and geographies would ultimately have a tool to have all the data available at a single point of reference. Whether it would finally mean a single person seated in his chair and controlling the entire world in consultation with his desktop/  laptop / handheld Watson only the time will tell. 

Ginni Rometty even goes as far as to say that cognitive computing will now be used for each and every decision that mankind will make henceforth.  It is estimated that approximately 80% of data generated by us is unstructured, and cognitive computing is the best resource that we have if we have to make decisions which can draw upon all the data which is available to us. Pretty soon it will be Watson telling us “Elementary my dear folks!”

Lazybee

3^rd September 2016