IBM's Watson: A Watershed Event For Information Technology--And For Society

On the surface, Watson, IBM's "Jeopardy"-playing computer system, is impressive in its ability to answer complex, ambiguous questions expressed in natural language. In fact, Watson will likely be a centerpiece for IBM's centennial celebration in June. But Watson is more than just an impressive computing system. In fact, in the future, after Watson's technology is used and deployed in multiple ways, we expect this week's Jeopardy matches will be seen as a historic event in information technology and big data and analytics that also portends much for society in general.

A "natural" language is any language spoken or written by humans, such as English, rather than a computer language, but there are areas where the two intersect. For example, a user performing a natural language search would enter a question in plain English rather than depending on keywords, such as those that trigger a Google search. Ideally, a true natural language query should return a single, very precise answer with a high degree of confidence rather than the many, many results a search engine delivers, leaving the user to determine what is relevant.

No disrespect to Google (whose original creative breakthrough has proven to be very important and valuable) or to other search engines like Microsoft's Bing, but solving the natural language problem is difficult. The problem is not just in syntactic analysis (computers have been able to parse sentences for some time), but rather in dealing with the subtleties, nuances and ambiguities common in natural language. Faux natural language attempts have failed; the best known is probably Ask Jeeves, which morphed into Ask.com and now is simply a front-end to a conventional search engine. It was a minor league attempt to a major league problem.

Some have suggested that artificial intelligence (AI) would offer a natural answer to the natural language question, and it has had its fits and starts over the years. But, even though AI can now point to a number of successes (primarily in well-defined domains), it has only provided partial solutions to the natural language problem.

IBM decided to take on natural language search as a Grand Challenge--an IBM R&D project that is technologically important and difficult, but whose success is easy for the average person to understand and meaningful for business and society. A past IBM Grand Challenge was the company's Deep Blue, a supercomputing system that defeated the human world chess champion Gary Kasparov in 1997.Now, with chess, the measure of success is easy: Simply beat the world chess champion in a match of several games. But what is the equivalent benchmark for natural language questions and answers? Well, as it turns out, the long-televised U.S.-based game show "Jeopardy" provides the benchmark. In "Jeopardy," three contestants have to answer a variety of difficult natural language-posed questions and answers.

Let's look at what Watson has to do. IBM states that there are five dimensions that the technology of automatic question answering has to take into account for a "Jeopardy" game: broad/open domain, complex language, high precision, accurate confidence and high speed:

In essence, "Jeopardy" has served two purposes for IBM. One is to provide a source of questions for testing Watson and a measure of its improvement over time by comparing its success to what is necessary to succeed on "Jeopardy." The second is to serve as a "final examination" to determine whether Watson can beat the two best players in the game's history, which is the equivalent of simultaneously beating a pair of world chess champions. The result of this examination is to use "Jeopardy" as a showcase to the average consumer of the power of big data, analytics and workload optimized systems.

Watson is a self-contained hardware system, meaning that all the computing power, storage and network connections are local. There is no connection to the Internet, so Watson cannot search for information that is not already available. Rather than the custom-built Deep Blue components IBM used in the Kasparov matches, for Watson, the company put together a lot of commercially available hardware.

In fact, Watson represents what IBM calls a "workload-optimized" system. That said, the sheer numbers are impressive: 90 IBM Power 750 servers with 2880 POWER7 cores, 500GBytes per second on-chip bandwidth, 10GByte Ethernet network, 15 terabytes of memory and 20 terabytes of clustered disk storage. Overall, the system operates at 80 teraflops, and, obviously, massive parallelism is essential.

Doesn't this mean that IBM is simply throwing hardware at a problem, a la Deep Blue? The answer is no. Hardware is a necessary component of a natural language system but is not entirely sufficient. If sheer speed alone were the issue, computers would have won a long time ago. (Many computers operate technically faster than the human brain. For understanding more about the brain versus the computer from a hardware and software perspective, see Jeff Hawkins' "On Intelligence" and Ray Kurzweil's "The Singularity Is Near"..Watson's secret sauce is IBM's Deep Question-Answering (QA), a technology the company describes as a massively parallel, probabilistic, evidence-based architecture. That is quite a mouthful, but, basically, a large number of natural language processing, information retrieval mechanisms, and machine learning and reasoning algorithms work together in concert. This is very much a deep analytics process and no single algorithm is sufficient. (Deep QA uses about 100 now.)

For any single question, hundreds of sources may be examined and thousands of possible answers generated. Potentially tens of thousands of pieces of evidence have to go through a hypothesis and evidence scoring process. A synthesis takes place (after all, massive parallelism is employed) that leads to final confidence merging and ranking in order to deliver a final answer with the greatest confidence possible.

No, Watson was not designed to pass the Turing test, and it is nowhere near sentient (at least, no one thinks so). Moreover, human intelligence still maintains preeminence in such domains as musical and artistic expression, as well as in imaginationl. Moreover, Watson is not the next big thing in the sense of a commercially available product that has everyone agog and generates billions of dollars of revenues. And it is not likely to be the next big thing on the Internet (although making Watson-like capabilities available online is a real possibility for IBM). So, if you can't buy it yet or use it yet, how can it affect your life?

IBM plans to turn to vertical industries, and a "Dr. Watson" for the health care field is likely to be the first commercial instance of the technology. The purpose would be to complement a physician by helping examine all the evidence (such as from medical journals and medical images) for a patient to form a broader, deeper perspective. For example, Dr. Watson could examine evidence outside of a specialist's area of knowledge that may apply to the patient. IBM believes that a Dr. Watson could save lives and that seems very likely the case.

In any event, IBM seems most comfortable in dealing with applications that are the extension of its enterprise-business focus. While that is fine, IBM may not exploit Watson's full potential. For example, it does not seem like Watson will appear on the Internet anytime soon in a Google-like service. This is understandable, since it is a different business model than IBM is used to. But given Watson's potential, why not license the technology to someone who can make use of it?There are other possible areas where Watson's ease of use could be valuable. Making Watson available via a cell phone would allow people to ask a natural language question and get back a very precise answer verbally (rather than having to try to review information on a small screen in text). Hint to IBM: This would be great for the blind, who could use it as a general-purpose application. Another option is the possibility of a micro-Watson on a desktop. A GUI provides a richer set of choices than command line, but can you really find the functionality that you need when you need it? Help features are notoriously unhelpful. Watson could be a godsend for these and other sticky areas.

Still why is Watson so important? What use is it? As Benjamin Franklin once said, "What is the use of a newborn baby?" If you did not know that Watson was a computer, could you tell it apart from the other contestants? The answer is no, and so Watson passes a very important intelligence test that some would not have thought could be passed for many more years, if ever. That alone qualifies it as a milestone in computer intelligence.

I will have to admit mixed emotions about whether I wanted one of the humans, Brad Rutter or Ken Jennings, or Watson to win the "Jeopardy" matches. I respect human intelligence, but, as a long-time member of the IT community in one way or other, I love to see advances in technology that benefit society. Never forget that Watson is still based in human creativity. However, the question is moot. IBM's Watson has demonstrated that it has solved the natural language problem from a question and answer perspective at a world-class level. Watson's dominant win is a milestone that will be remembered in information technology history.

It is rare to have a technology that once appeared only in science fiction come true. The cell phone is an instantiation of the famous Dick Tracy wrist radio. Watson is an instantiation of the Q&A computer on the original Star Trek television series. Though the derivative benefits of Watson are some time away from being realized, even though those benefits are subtler than something we can touch (a physical product) or see (an application on the Internet), they are going to be real. Hey, it may be your life that Dr. Watson helps to save. From a societal perspective, it could be argued that IBM has the obligation to see that such a unique and potentially beneficial technology is used as broadly as possible, even if this means licensing or making it open.

I feel impelled to give out some thank yous. We constantly hear about the need for innovation as essential to our society. With the diminishing resources for basic research (witness Bell Labs), the seeds for technological progress are hard to find. Fortunately, some large IT vendors continue to invest in and perform very good and valuable research that goes beyond the next revision of a current product. IBM is one of those companies.Watson has created new options and opportunities in a very important area. The "Jeopardy" organization--and that extends to the contestants--should be commended for providing a fair and challenging platform for natural language computing. IBM, from the top down, should be commended for investing in Watson. Few if any other IT vendors would be willing to invest in something that might not work and, even if it did work, would not likely have a measurable short-term ROI. In Watson, IBM has a centerpiece for its centennial and a memorably innovative start for its second century.

IBM is a client of David Hill and the Mesabi Group.