IBM Watson is one of the most impressive technologies released in the last few years. Focused on enabling cognitive and artificial intelligence(AI) capabilities, Watson has become the cornerstone of IBM’s strategy for the next decade. Big Blue has been pouring billions into the development and commercialization of Watson and has made it a key piece of some of other important solutions such as the Watson IOT Platform.
Functionally, Watson enables a series of APIs that abstract cognitive capabilities such as speech, text, vision, knowledge and data analysis. These APIs are exposed via the Watson Developer Cloud(WDC) which is a combination of home ground Watson technologies and smart IBM acquisitions such as Alchemy.
Watson’s Competitors are Catching Up
While IBM Watson remains the undisputed leader in the cognitive computing space, some of its rivals have been delivering interesting technologies that could become relevant competitors in the near future. The early success of Watson and the rapid growth of the AI market has made companies such as Google and Microsoft to enter the space with really innovate offerings.
During the past couple of years, Microsoft Research incubated a cognitive platform under the name Project Oxford. A few months ago, Oxford became generally available as Microsoft Cognitive Services which has become the AI and cognitive computing arm of Microsoft’s Cortana Intelligence suite. Similarly, Google has been releasing different AI capabilities for the last decade. However, Google’s AI efforts have redoubled recently with the release of platforms such as the Natural Language Processing, Vision and Speech APIs. Microsoft and Google’s decades of experience building highly scalable, mission-critical system can certainly make them contenders in the cognitive computing space.
While Microsoft and Google have been very public about their AI platforms, Amazon has remained somewhat quiet in terms of its AI initiatives. However, we are already seeing flavors of Amazon’s AI capabilities in the form of the Alexa platform that powers Amazon’s Echo assistant. AWS’s large developer, customer and partner ecosystems will be a strong factor pushing Amazon’s AI initiatives.
In terms of functional capabilities, Microsoft Cognitive Services and Google AI APIs look very similar to Watson. IBM’s advantage in the market has been Watson’s early adoption, market traction and impressive revenue numbers. With rivals rapidly closing the gap, we can expect IBM to enhance Watson with new and differentiated capabilities.
Five Capabilities that Could be Included in Watson’s Roadmap
The roadmap of the Watson platform must be incredibly ambitious if it intends to continue its dominance in the highly competitive AI market. Based on the current state of the AI ecosystem, I’ve outlined a few ideas that could be included in the short term roadmap of the Watson platform.
Google is doing a remarkable job creating industry solutions based on the DeepMind AI platform. Microsoft has also been successful penetrating different industries with its Cognitive Services solution. Being the market leader in the space, IBM has a unique opportunity to deliver a new generation of industry-specific solutions powered by the Watson platform. While IBM has already made some progress in this area, it should be a strong segment of focus in the near future. Verticalizing Watson in the form of industry-specific solutions can result in a very unique competitive advantage in the next phase of the AI market.
New Cognitive Capabilities
New cognitive areas such as handwriting recognition or video analytics could be excellent additions to the Watson platform. As Watson expands, it is important that continues ahead of its immediate competitors exploring new cognitive capabilities.
Server Side Cognitive Programs
Today, applications interact with Watson via APIs. Any business logic about what actions to take based on the results of a cognitive algorithm live within the client application. That form of interaction highly contrasts with the way the human brain works. Neurons in the brain’s neocortex process information from different sensors [vision, speech, audio], recognizes appropriate patterns and take actions accordingly. In the near future, Watson can evolve to enable custom server side programs that combine different cognitive capabilities of the Watson platform. Currently, the combination of Watson and OpenWhisk almost accomplishes this goal but it would be nice if the next version of the Watson platform included these capabilities as part of the default feature set.
More Knowledge APIs
The current group of Watson knowledge APIs it leverages Wikipedia almost as the exclusive data source. Expanding knowledge APIs with other sources [ex: WebMD, Library of Congress, etc] on different verticals would enrich Watson’s applicability on different domains.
Model training remains one of the big challenges of the Watson platform and other competitive cognitive platforms. Although IBM has released a couple of tools for training Watson, they are pretty basic and fairly expensive for mainstream adoption. Improving the current toolset for training Watson models should be part of the short-term roadmap of the Watson platform.