For an in-depth look at where Watson is headed, we spoke with Carlos Paez, Lead Mobile Developer, MobileFirst Global Center of Competence.
Where does Watson stand today?
At this time IBM is concerned with building an ecosystem around Watson. This is the inspiration for the Watson Mobile Challenge. IBM will make the APIs for Watson available to an increasing number of parties, starting in Q2 2014. Paez refers to this as “opening the sandbox” to create a strong developer community.
Can Watson capabilities be embedded in devices?
Paez states that some capabilities will be available in devices, however, for the most part Watson will be offered through services. He states that there are many corpuses of knowledge and that they can be accessed through standard data connections, with the ability to scale capabilities in the server.
Can Watson play a backend role, for example, in apps development?
Paez responds that the direction of mobile apps is towards putting more things in the cloud. Watson can store a lot of data on servers and offer an API for access. The APIs would be available for all types of operating systems.
How much customization or preparation (in terms of questions, hypothesis, scoring) does it take to make Watson workable for a business?
Paez responds that Watson requires some training. However, he states that this would be a one-time event. He points out that the amount of information Watson can process on its own is already impressive and there are general behavior patterns that Watson applies that cut across all types and fields of data.
What are future plans for enhancing Watson’s broad capabilities?
The current version of Watson, perhaps it can be called Watson 1.0, is not the end of development by any means. The next version, which is being worked on will include the ability to ingest images, video and other input. Paez states that “natural language processing is great, but it is only part of the picture.” He notes that capabilities such as “sentiment analysis” – gleaned from tone of voice or speech pattern – is something Watson can start to do now.
Does Watson have particular applicability to specific verticals, the “low-hanging fruit”?
Paez mentions consumer shopping, providing the example of a smart assistant to help an individual plan a trip, perhaps to hike Kilimanjaro. It can produce all types of information about the destination and help with planning. He also refers to Watson’s ability to ingest personal information, so that it can take into account in its responses, for example, which vendors the individual has purchased items such as clothing from in the past.
What’s really of key importance for the development of Watson?
Paez emphasizes the collection of data. We at mobilecloudera.com have written extensively about how the mobile cloud enables vast applications where “big data” is collected from the “edge”. (See, for example, Premise: Hyperdata Gathering From the Edge and the Mobile Cloud 1/29/14.) Paez responds, “That’s exactly right; it is why ‘mobile’ is in the name of the Challenge.” Data can be captured by smartphones and input to Watson. Often unstructured data can be gathered and Watson can come out with a structured answer.
Does the Challenge reflect a marketing-oriented focus?
(We noted that from attempting to fill out the entrants’ application form.) Paez agrees that the emphasis is on building a suite of products and services. These will be technologically agnostic and accessible through multiple platforms, so that the user’s options are not platform-limited. He states, “The technology will come after the fact.”
How will IBM commercialize Watson, for example, what about pricing?
It’s too soon to really explore how IBM is going to commercialize Watson. When asked about potential pricing, Paez states in a general way, IBM could use models such as Google’s, which provides access to some cloud services and APIs free, up to a certain level of use, with charges for additional usage. Also they will charge if someone builds a commercial service on Watson.
What if storage pricing goes to virtual zero?
Paez acknowledges that this is a possibility, but states the IBM will use other ways to charge. He mentions charges for data, or for connection to a service or for managing an API.
The Watson Challenge is only the beginning phase of IBM’s go to market strategy around mobile-centric cognitive computing services which rely on big data.