After defeating the two biggest Jeopardy champions of all time, people are buzzing about IBM Watson. Reactions range from excitement over the possibility it represents, to fear over the potential impact on society. Count me among those who believe that this technology will have a positive impact on business, Government and humankind. With every new game changing innovation, there is always fear over the potential impact on humanity. However, a machine will never have wisdom. It will never have the ability to invent, have empathy, or match the knowledge and ability to reason of a smart, experienced human being. But this form of advanced analytics will help us perform better.
With the attention given to IBM over the last month, it’s easy to think that IBM Watson is the discussion. Well it’s not. It’s about advancements in analytics, computing power, and the explosion of data that allows us to deliver new capabilities to solve business problems, address government challenges, fight crime, and improve healthcare. The possibilities are endless.
IBM is already talking about developing a physicians’ assistant service that collects a patient’s health information and analyzes it for medical diagnosis. Other health care applications include a capability to automatically identify and flag anomalies on MRIs and other images that a radiologist may miss. Health care isn’t the only place where advanced analytic applications will create value. I believe IBM has shined the light on the analysis and correlation of large volumes of data and the possibilities it represents for Industry, the legal profession, the intelligence community, law enforcement, customer service executives, city planners, etc. IBM provided several post-Jeopardy examples of advanced analytic applications including:
- A monitoring system that analyzes a county’s historical and real-time traffic information while tapping data sources on where accidents occur most, where the most speeding tickets are given, and where the public transportation routes are. The system could then be queried by a city planner to answer questions such as: Where should we add more traffic lights? Where should we raise or lower speed limits? Where should we add a carpool lane? What should we do to make traffic safer and more efficient?
- An automated email or IM-based customer service system where a customer could ask any number of questions in natural language and theoretically get quick and accurate answers from a machine drawing from an array of data sources in real time.
Another significant part of this story that IBM Watson effectively demonstrated is that we can analyze data, extract insight, and provide answers in real time. Overnight batch mode could be a thing of the past, with answers and confidence levels available at the point of decision.
As promising as all this sounds, it is not as easy as what we saw on Jeopardy. It took a long time to train Watson to perform at a winning level. I’ve experienced this first hand through the iterative refinement that makes a fraud mitigation solution better at detecting fraudulent claims. By focusing on natural language as well as structured data, fraud is better identified. But you must capture the knowledge of humans, apply machine learning techniques, and refine models over time through the iterative analysis of claims. This refinement process – much like the process utilized to win at Jeopardy – enables these models to quickly identify fraud while significantly reducing false positives and negatives. You can read this Forrester Blog for an additional perspective on the challenges faced when deploying “Watson-Like” solutions.
Nevertheless, I believe the future holds much promise in supporting decisions at the point of interaction – whether that interaction is with a patient, someone calling into the call center, a law enforcement official, or a city planner.
What the IBM Watson experience has done for me is provide a platform to discuss real-world examples of how these capabilities will be applied in the future.
For additional insight on what lies ahead, I recommend the articles below.