What IBM’s Watson Can’t Tell You

As sophisticated as artificial intelligence is getting, computers still can't interpret what's happening in the real world—unless auto-ID systems feed them the data they need.
Published: October 21, 2015

IBM has been running a series of ads promoting its Watson computer system, which uses artificial intelligence to learn and is capable of examining vast amounts of data—in one ad, Watson says it can read 800 million pages a second—and interpreting the information. I love the ads, and I love the way this power is being put to use in some industries. Watson, for example, has been looking at MRIs and other types of medical images—as well as the results of past treatments—and recommending the best treatments, which is incredible.

But Watson, like all computers, is limited to analyzing the text, numbers, images and other data humans feed it. Watson does apparently have a camera enabling it to perform some facial recognition—in another ad, for example, it recognizes Ken Jennings, a contestant whom the computer faced on Jeopardy! in 2011.

Beyond that, this highly sophisticated computer can’t determine what is going on in the real world and recommend ways to improve efficiencies or optimize processes. For that to happen, companies must deploy automatic-identification technologies that accurately identify and locate objects and individuals in the real world, and then feed that information to Watson. The benefits would be enormous.

Take retailing, for example. Every major store chain in the world would like to have Watson analyze massive amounts of data, including tweets about their products, and make recommendations—for instance, “People liked your product but they didn’t buy it because it wasn’t on the store shelf in their size and the color they wanted.” But Watson can only tell you this if it has real-time information about your inventory—and if your inventory data is accurate.

One application being developed for Watson is the ability to analyze hospital purchases to improve procurement. I’m sure some incremental improvements can be made, which would save hospitals money. But imagine what Watson could recommend if it knew where and when a hospital’s mobile medical equipment—gurneys, oxygen pumps, wheelchairs and so on—were being used, and how many devices were sitting around waiting for repair or maintenance. Similarly, if a facility could automatically collect data regarding the consumption of drugs and medical devices, Watson would not have to rely on faulty information (people often record their usage inaccurately) to optimize the purchase of these items.

Enabling Watson to experience the physical—not just the digital—world opens up a whole new realm of efficiencies. That ability is possible today. All it takes is for companies to collect data automatically via radio frequency identification, video and other technologies, and to then feed that information into computers like Watson that can analyze what is happening in the world and suggest ways to improve efficiencies and optimize processes. Given the amount of waste that exists due to our inability to identify, track and manage everything in the real world with IT systems, that’s going to be an exciting leap forward for businesses.

Mark Roberti is the founder and editor of RFID Journal. If you would like to comment on this article, click on the link below. To read more of Mark’s opinions, visit the RFID Journal Blog, the Editor’s Note archive or RFID Connect.