IBM Watson Makes Things Elementary, Indeed

The cognitive computing system makes for an ideal sidekick—in museums, kitchens, hospitals and classrooms

(Christian Charisius/dpa/Corbis)
smithsonian.com

Museum audio guides are handy, to a point. The tours might explain a particular brushstroke used in a painting or the origins of a striking whale artifact, but because they are automated and preprogrammed, they aren't able to answer a user's specific questions or play to a person's interests.

IBM's India-based research team is harnessing the smarts of Watson, the company's cognitive computing system, to create the ultimate tour guide. The engineers are developing Usher, a mobile app that comes loaded with facts and figures about the different items showcased at museums worldwide and quickly offers up Siri-like answers to queries about everything from the Pleistocene to Jackson Pollock. Still in a trial phase, the app has only been demonstrated with a limited data set to showcase its functionality. Within the next year, its developers aim to identify the museums it will target first.

What's particularly unique about Usher is that it responds to the user's behavior. Leveraging geolocation functions on the smartphone, including the accelerometer, gyroscope and compass, Usher can keep up with the speed and location of the user, answering questions about nearby works and providing unprompted information in real time. By sensing where in the museum an individual is, the app can offer extremely precise facts. Walk by The Red Vineyard, for instance, and Usher might say, "The painting on your left is supposedly the only piece Vincent van Gogh sold in his lifetime." It also informs the user about an artwork's style and when it was created. Based on the questions posed to it, Usher can get to know a user's interests and provide more insight into those areas. Deemed an "intelligent tour companion" by IBM, the app aims to serve as a curator of sorts once it's completed.

As such, rather than act solely as a static resource, Usher takes an active role in the user's museum experience. This level of active engagement is emblematic of the power that Watson has to not only provide answers, but also offer informed and nuanced guidance, based on the data it has absorbed.

Marc Teerlink speaks at Smithsonian magazine's 2015 Future Is Here Festival

Its core strength, Watson has an astounding ability to analyze large quantities of information and distill it down to useful bits and pieces. "Seven seconds is all it takes for IBM Watson to read 49 million bookpages or medical files," said Marc Teerlink, IBM Watson Group's chief business strategist at Smithsonian magazine's recent "Future is Here" festival.

The cognitive computing system made a splashy debut as a contestant on Jeopardy in 2011, besting human incumbent Ken Jennings in a runaway victory. Its ability to "go through an obscene amount of data," Teerlink said, a volume no one person could parse on his or her own, made it the ideal contestant for the game show, which requires an expansive knowledge base. That same skill makes Watson an incredibly useful search engine for an array of fields, from museum education to medicine.

But Watson is much more than just a search engine. The technology powering the system allows it to look at the "layers of meaning beneath the surface" of the information it's examining, or rather, to think like a human and find connections between the topics at hand. In a search about "concussions," for example, it would be able to link the results found with related topics like "football" and "brain injury." At its most sophisticated, Watson offers guidance, so rather than a flat answer to a question, it provides the context behind its response and an explanation of why its recommendation serves as a fitting solution.

The system also learns from experience, constantly building on its reservoir of knowledge. This spring, IBM and the Institute of Culinary Education coauthored a new cookbook, Cognitive Cooking with Chef Watson. For it, Watson was able to develop more than 65 interesting and experimental new recipes based on the recipes of top chefs it had analyzed at the Institute of Culinary Education. After reviewing the data, it learned what ingredients pair well together and was able to suggest unconventional combinations, including Indian Turmeric Paella and Creole Shrimp-Lamb Dumplings.

The startup Elemental Path is using the technology to power its CogniToy, an adorable, miniature green dinosaur that serves as an accessible encyclopedia, able to answer the many questions kids ask. "How far is the moon?" asks a child in a recent Kickstarter campaign video for the toy. And the dino replies, "The moon is about 250,000 miles from Earth." The toy, in turn, asks the kids questions, learns their interests and adjusts its responses and lessons to a child's preferences. If a child enjoys music, CogniToy can customize future vocabulary lessons to include more music terms.

"Watson can be your helper, your sidekick, your assistant, augmenting what you do," said Teerlink.

In a current partnership with Memorial Sloan-Kettering, Watson is being used as a support system for oncologists, by helping physicans look at historic data on patient patterns and identify the best treatment possibilities. With 700,000 new scientific articles published a year, it can be challenging for physicans to keep up with the latest medical research. Watson can read through the latest papers quickly, convey key insights and make recommendations to doctors. In a similar application, Watson helps connect Mayo Clinic patients with clinical trials, in a matter of seconds as opposed to weeks.

Today, IBM offers four key means of working with the technology behind Watson. There is an OpenAPI available on IBM's cloud innovation platform Bluemix that developers can experiment with and utilize for apps and other tools they are interested in creating. The startup Welltok, for example, has integrated Watson into an app called CafeWell Concierge, which offers personalized recommendations on how to eat and live healthy, with nearby food suggestions and reminders to schedule doctor's appointments. 

Once groups have built a tool that they are interested in scaling, much like the CogniToy, IBM has an open application for partnership. By partnering with Watson, companies gain access to an IBM support team and additional resources to help them develop their product. So far, in its first year of the program, IBM has welcomed 270 commercial partners into its ecosystem, from healthcare companies like Johnson & Johnson to music discovery platforms like MusicGeek by Decibel.

Additionally, IBM plans to continue a series of competitions and hackathons at college campuses this year, including the second annual Watson University Competition, challenging students to develop new applications of its technology. A student team from the University of Texas at Austin placed first in the inaugural contest with CallScout, an app that provides information about local social services to Texas residents.

IBM has also developed package solutions utilizing Watson technology that organizations can purchase and integrate into their existing infrastructure, providing support in key areas including engagement, discovery and decision-making. Watson package solutions help companies comb through data sets to retrieve answers for customers' questions. The United Services Automobile Association (USAA), an insurance company focused on serving members of the military, uses Watson to search through information about services for those who are part of the armed forces, leveraging the tool to power the USAA website and inform customer service. Companies can purchase these packages to help supplement different elements of their business.

Watson's overwhelming power lies in its ability to help users sift through legions of data to identify the answer they need—and there is an infinite number of use cases for it.

"Why can't we get an overview of all the knowledge collected in the past 5,000 years," asked Teerlink, "and use it today?"

Tags

Comment on this Story

comments powered by Disqus