For Jeopardy fans like myself, the last two nights of play have been quite different than the traditional matches. Not just because the all time champs Ken Jennings and Brad Rutter have returned, but more because they are playing against a computer named Watson. Watson can “read” questions, interpret meanings, find natural language relationships and answer questions with a precise confidence interval… all in the form of a question, of course. So what if Watson ran your supply chain?
“Boring.” “What a waste of money.” “No computer will ever be smarter than a person.” These were some of the comments posted under a story I just read about Watson. And, while watching a computer rapidly answer questions with incredible precision while its human competitors stared off in frustration doesn’t make the best television, Watson is far from “boring.” What is so fascinating about Watson’s capability is that it moves way beyond the search capability we’re currently used to. Watson understands analogies, can find relationships in otherwise disparate topics and can rate the likelihood of being correct… all within seconds. The implications for supply chain analytics are HUGE.
We’ve recently published several reports on demand sensing, pattern based strategy and social media for supply chain (http://blogs.gartner.com/matthew-davis/2010/10/04/social-media-to-improve-supply-chain/) that hinge on being able to identify patterns. Patterns in demand, customer preference, macro-economic factors, etc… For example, many companies maintain an active social media presence to open a dialogue with customers on product features, trends in user communities and other “like” products that they’re interested in. The problem with this data is that it comes from many sources, is in written “natural” language and is extremely difficult to group for meaningful analysis.
In theory, Watson’s capabilities could be used to rapidly analyze all of that available data to decipher out meaningful analytics to guide strategy. In addition, it could look for relationships within and across all of these disparate data sources to identify patterns.
As they described Watson’s logic engine, IBM used an analogy of a doctor trying to make a diagnosis to describe Watson’s potential. A doctor might have a hunch as to an expected problem, but has many different factors to consider. Perhaps some even unknown. Add onto that complexity with the reality that most medical records and notes are still hand-written and the ability to compare all of one patient’s data with the massive amount of data online seems impossible. But what if the doctor could load the patient’s data into Watson which would then use its massive relational logic engine to find patterns in the data as a check of the doctor’s hypothesis? Science fiction? Maybe today… but you can clearly see the possibility as Watson fires away at questions that today’s search engine could never handle.
Now think supply chain… Maybe Watson reviews buzz on Facebook about new product launches of certain electronics and is able to tell an accessories provider to expect a huge spike in demand. Or maybe Watson recognizes that when unemployment levels hit 10%, your company’s demand tends to fall by 4%. When you consider the incredible amount of data available online, including all of the active social communities adding content daily, the possibilities are quite exciting.
Now, all of this said, it took so many servers to support Watson that they had to be housed offsite. And Watson did have a couple incredible misses that seemed out of the blue. BUT, the results speak for themselves so far. At the end of the first round of play, Watson leads both players by over $20,000. Watson’s short-term fate on Jeopardy will be decided this evening, but the long-term implications are enthralling.
What do you think? Will computers ever be able to understand all the intricacies of human speech? If so, when? Do you see applications for your supply chain? Or is this all just a publicity stunt?
Comments or opinions expressed on this blog are those of the individual contributors only, and do not necessarily represent the views of Gartner, Inc. or its management. Readers may copy and redistribute blog postings on other blogs, or otherwise for private, non-commercial or journalistic purposes, with attribution to Gartner. This content may not be used for any other purposes in any other formats or media. The content on this blog is provided on an "as-is" basis. Gartner shall not be liable for any damages whatsoever arising out of the content or use of this blog.