IBM has been confident its machine-learning project, Watson, can do much more than just win at "Jeopardy." To show off Watson's power, Big Blue set up a contest last February for mobile developers to produce apps that make the most creative and inventive use of Watson's cognition.
Now, after three rounds of eliminations, the top three mobile app submissions are in, and each of the winning apps deals with an implementation of machine learning that's either already proven itself or shows promise for doing so.
The first and perhaps most obvious role for Watson is in health care. This is represented by the GenieMD app, which "empower[s] individuals to take a more active role in managing their health by delivering a holistic view, making health data actionable and shareable," according to IBM.
"When a customer has a health question," explained GenieMD's creators, "GenieMD rephrases the question based on the medical conditions, medications, allergies, etc. and forwards the modified question to Watson to receive a targeted and specific answer."
GenieMD is not itself new -- version 6 of the app for iOS was released back in March -- but the Watson integration is. Adding a predictive component to an existing, entrenched health care product makes even more sense, especially since a strong part of Watson's existing track record revolves around health care.
Education is another area under consideration for Watson's powers, and Majestyk Apps' Fang (Friendly Anthropomorphic Networked Genome) hints at how this might work. It sounds like an AI-powered version of the old Teddy Ruxpin toy: "An artificially intelligent, cuddly plush companion for children capable of answering questions posed in natural language, learning from natural dialog, and interacting and reasoning with its human counterpart."
The idea is that as the child grows, so do the experiences with Fang -- which can then be paired with a tablet or phone as a way to gracefully outgrow the toy experience and move to another, more age-appropriate medium.
The third app, by Red Ant, is in some ways the most conventional. It's designed to train retail sales personnel, and it uses Watson's APIs to do deep analysis of customer preferences, both based on their previous buying histories and their overall demographics. "Watson will provide customized selling points unique to that customer," explained Red Ant, "either on screen or via text-to-speech on an earpiece (and someday via wearables like glasses or smart watches)."
This squares with the most obvious and familiar business-case uses of Watson. After all, analyzing trends and deep-mining behaviors are exactly the tasks IBM has been expected to monetize most readily with the service.
Health and fitness, though, seem to have captured most of the attention of both IBM and the Watson contest developers. Of the other 25 finalists for the contest, eight total were either health or health-related. That's not much of a surprise given the rising interest in wearable tech and the "quantified self," disparate as those initiatives currently are.
Has new ground been broken here, with or without Watson? It's tough to tell this early on, especially without a clear real-world idea of how Watson will make these apps more worthy than their non-Watson equivalents. It's also still unclear how IBM plans to monetize Watson as a service, whether by charging on a per-app, per-API-call, or per-compute-unit-hour basis.
Consequently, the big issue in the long run isn't whether these apps take off -- it's whether Watson's full potential can be realized through consumerized apps and how that will spur more direct competition in machine learning as a service.
This story, "3 mobile apps bring home IBM Watson's power," was originally published at InfoWorld.com. Get the first word on what the important tech news really means with the InfoWorld Tech Watch blog. For the latest developments in business technology news, follow InfoWorld.com on Twitter.