Wednesday, March 7, 2012

Watson computer - the INTELLIGENT one !!!

The human-like intelligence- ARTIFICIAL INTELLIGENCE !!

Can a machine be intelligent ? Can it talk to people in their natural language ? Can it read our lips ? Is it possible to build a machine as smart as a person ? Can the human intelligence be replicated ? Can we build a person ? or does it only confine to fiction ?  The question has been floating for decades but without a concrete answer !!  The computer has always been the king when it comes to processing huge amount of data/information.The immense rate at which it has grown inspires the confidence. The confidence is high enough to at least inspire a lot of characters in fiction.  The concept looks mesmerizing but no computer/robot has been able to interact with human as seamlessly as hollywood imagines. The problem is - the human brain, a very complex entity that defies any attempt at replication. Our brain is immensely sophisticated. The simple skills that humans master early in life like understanding language or recognizing objects continues to baffle the researchers.

The quest for artificial intelligence has gone on for decades. There is a huge difference between data and information. Computers process data but can they extract information out of it ? Can a machine possess knowledge ? Can it learn from experiences ? These are questions that decide if a system is intelligent. Lately there has been a lot of research in the field of artificial intelligence and machine learning. But no machine can understand natural language and converse with a human for it includes puns, irony, riddles and double meaning sentences. There has been a lot of work to achieve it and this article looks at one such attempt. An attempt by IBM.

The Chess Playing Computer:

You might have heard of Deep Blue- the chess playing computer !! It was an initiative by IBM. If wikipedia is to be believed, On May 11, 1997, the machine won a six-game match by two wins to one with three draws against a world champion in chess- Garry Kasparov. The system derived its playing strength mainly out of ts brute force computing power. It was capable of evaluating 200 million positions per second. Computing at an extremely high speed it could evaluate multiple steps and search the step that would result in maximum chances of winning or minimum chances of loosing. There were adverseries who argued that the rules of chess are very well defined. So it can easily be coded in a computer system and this in no way means that the system was intelligent. Thanks to these people IBM came with a better system- The Watson Computer!!

The motivation Behind the Watson computer:-

Its an interesting story that led to the thought of designing this system. Watson computer was designed to play Jeopardy!. Jeopardy is a unique quiz-show in which there are three contestants. They are given clues leading to an answer and the contestants are required to guess what the question would be !! Sounds kind of interesting. Right ?? The show features trivia in history, literature, the arts, pop culture, science, sports, geography, wordplay, and all other topics. To know more about Jeoprady! you may visit the wiki article or watch a television show of it on youtube.Getting back to Watson,since Deep Blue's victory over Garry Kasparov in 1997, IBM had been on the hunt for a new challenge. Once Charles Lickel went out for a dinner with his coworkers. There he noticed that the restaurant they were in had fallen silent. The cause of this was the airing of a show on television featuring: Ken Jennings, who was then in the middle of his successful 74-game run on Jeopardy!. He thought of quiz show as a possible challenge for IBM, and passed the idea on. In 2005, IBM Research executive Paul Horn backed Lickel up and David Ferrucci took the challenge.

What is Watson ? What is it capable of ???

According to IBM, Watson is an efficient analytical engine that pulls many sources of data together in real-time, discovers an insight, and deciphers a degree of confidence. Lets analyse this definition. You can pose questions/hints and watson will search the most relevant things for you. Is that not what the search engines do ? The point is, watson takes you a step further. Watson can understand the actual meaning behind words, distinguish between relevant and irrelevant content, and ultimately demonstrate confidence to deliver precise final answers. As a result of its deeper understanding of natural language, it can process and answer more complex questions that include puns, irony and riddles common in natural language. Look at the words in bold. Yes, thats exactly how it differs from your favourite search engine. Ask watson- "What is the capital of Russia?” and it will speak out - “Moscow,” . As simple as that !! Now think of your search engines. Ask them the same- What is the capital of Russia?” It’ll award you with links to websites where you might find your answer.You then revise your search terms until you find a satisfactory answer. Woludnt it be nice if you could just ask the question, and get an actual answer rather than a list of documents or websites? Thats watson !!!

Watson- the Jeopardy Player:

Is Watson an improved search engine ??  No, it was never meant to be one. In 2011 Watson did what it was supposed to, it played the game of Jeopardy! with Ken Jennings and Brad Rutter, two of the most successful contestants on the show and emerged the winner too. Watson learns through machine learning, it learns from its mistakes, for example, initially it used to come up with very silly answers- some of which one can never expect out of a human. As part of the preparation, IBM constructed a mock set in a conference room at one of its technology sites to model the one used on Jeopardy! Human players, including former Jeopardy! contestants, also participated in mock games against Watson. This helped it learn a lot about the game. Due to its limitations with audio/video understanding, Watson received the clues as electronic texts at the same moment they were made visible to the human players. It would then parse the clues into different keywords and sentence fragments in order to find statistically related phrases.Watson's main innovation was not in the creation of a new algorithm for this operation but rather its ability to quickly execute thousands of proven language analysis algorithms simultaneously to find the correct answer.The more algorithms that find the same answer independently the more likely Watson is to be correct. Once Watson has a small number of potential solutions it is able to check against its database to ascertain whether the solution makes sense. Part of the system used to win the Jeopardy! contest was the electronic circuitry that receives the "ready" signal and then examined whether Watson's confidence level was great enough to activate the buzzer. Given the speed of this circuitry compared to the speed of human reaction times, Watson's reaction time was faster than the human contestants except when the human anticipated (instead of reacted to) the ready signal. After signaling, Watson speaks with an electronic voice and gives the responses in Jeopardy!'s question format. Watson's voice was synthesized from recordings that actor Jeff Woodman made for an IBM text-to-speech program in 2004. You would think is it only for a game that IBM spent all its time and money ? No, Jeopardy! was only a showcase of IBM's work in the field of AI.

The Science Behind an Answer:

The science behind formulation of answers by Watson can be thought of as a 4-step process:

  1. Question Analysis: Watson parses the question into parts of speech and identify the different roles words and phrases are playing. At this stage Watson has no idea of what the question can be. So it increases its chances of reaching the right one by considering many different options about what the question may be.

  2. Hypothesis Generation: For each interpretation of the question, Watson goes through millions of documents. At this point quantity is important as it does not want to miss out on reaching the correct reference. If it does, it can never chose the correct answer!!

  3. Evidence Scoring: It is not enough for watson to just come up with an answers but it has to support them with evidences and logic. Thousands of algorithms run in parallel to evaluate the score for each possible answer.

  4. Final Merging and Ranking: The scores are then used to rank the possible answers and evaluate the confidence watson possesses in its highest ranked answers. Note that the threshold confidence level based on which Watson decides whether it should give an answer or not is not fixed but depends on the scores of its opponents and the situation in the game.

What future has in store for Watson: 

Watson cant understand the love of a mother, the feel of music, the emotions but his enormous knowledge base, his skill to understand and interpret the natural language and his ability to learn are definite signals to a marked achievement in the field of technology. Watson is impressively intelligent. Beyond Jeopardy!, the IBM team is working to deploy this technology across industries such as healthcare, finance and customer service. In the words of Dr. David Gondek form IBM:
U think of internet. No one knew that internet would become such an important part of our life. Watson has a similar promise. It is transformative and you dont even know what it can become.

The whole project was not about Jeopardy! , it was about doing research, its about doing deep analytics and natural language understandings. Its about applying the technology to solve problems people really care about. According to IBM, "The goal is 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." Nuance Communications Inc has partnered IBM in the research project to develop a commercial product that will exploit Watson’s capabilities as a clinical decision support system to aid the diagnosis and treatment of patients. Physicians at Columbia University are helping identify critical issues in the practice of medicine where the Watson technology may be able to contribute and physicians at the University of Maryland are working to identify the best way that a technology like Watson could interact with medical practitioners to provide the maximum assistance. It has also been suggested by Robert C. Weber, IBM's general counsel, that Watson may be used for legal research. Just the same way Watson analyzed massive data in Jeopardy! to reach a set of hypotheses and list down couple of most likely outcomes, it could help Doctors in diagnosing patients. Watson could analyze the patient's specific symptoms, medical history, hereditary history and synthesize that data along with the mass of all unstructured and structured medical information in the world including (but not limited to) medical records of similar situations available to it. Watson uses this information in addition to the wealth of fundamental medical knowledge available in the form of all published medical books and articles fed into Watson. IBM had made it clear that Watson did not intend to replace doctors, but assist them to avoid medical errors and sharpen medical diagnosis with the help of its advanced analytics technology. IBM intends to use Watson in other information intensive fields as well, like telecom, financial services, government etc. To end with, I would say this is not the end of the list, but the beginning of an era and

Now, do you think its the Golden Age of Technology we are living in?



  1. Nice post..........a new thing I learned today.......:)

  2. Thanks shashank !! Its this sort of encouragement that keeps the enthusiasm going :)