Summary Bullets:
- Just one short year after an internal reorganization to more fully meld artificial intelligence (AI) with data and analytics, IBM is back with a new, more accessible vision for IBM Watson.
- This time around, the company isn’t focused on game shows or scientific discovery but instead on solving very basic, often human-centric challenges.
When it comes to chasing the market’s heated but somewhat unrequited love affair with AI, IBM has certainly done its part in terms of generating hype for its multi-billion dollar investment in IBM Watson. That hype, which has taken aim at some rather lofty goals such as identifying and diagnosing cancer, has not fully panned out, with some early adopters scaling back or halting operations altogether due to concerns over cost and efficacy.
This has given AI rivals Microsoft, Google, Amazon, and others ample room to operate with a more modest, more general scope of concern, namely in delivering highly operationalized, horizontal AI tools that speed development efforts and lessen the demand for data science expertise.
Recognizing that it has perhaps overshot the AI value proposition at least in terms of hype vs. reality, IBM has both drawn back from its earlier, over-aggressive use of the Watson brand and rejiggered its internal organization to better blend AI with the data that drives it. Now, the company is back with a new marketing message for IBM Watson, ‘Watson Anywhere,’ which intends to meet rivals head on in promoting the operationalization of AI for broad adoption.
The company put this new message on display at its annual IBM Data and AI Forum in Miami, Florida last week, driving home the message that IBM Watson is no longer to be considered a magic, anthropomorphic black box capable of winning Jeopardy, discovering new drugs, or suggesting your next favorite beer. (Note that none of these are made up.) Key to this new worldview is a much clearer set of swimlanes for Watson itself, which the company now breaks down into three parts.
- Tools (libraries, development environments, etc.) customers can use to build their own AI solutions
- Pre-packaged AI solutions targeting specific AI outcomes
- Embedded AI ‘features’ pervading IBM’s broader software portfolio
Permeating these three means of consuming IBM Watson, customers can now find a likewise clear-cut set of priorities which take direct aim at what we would agree are the top challenges faced by enterprise customers seeking to leverage AI in the real world, namely waning trust (in AI outcomes), unsuitable data, and inadequate skills.
These are truly laudable goals, but are they truly achievable beyond a marketing slick? As befitting a company accustomed to solving problems through engineering, IBM believes so and hopes to meet these challenges head on with software.
To this end and focusing on the problem of unsuitable data, this summer, the company released a new feature within IBM Watson Studio (a tool used to build AI solutions) called AutoAI, which quite literally automates many of the steps required to collect, prepare, and pre-process data for use within machine learning (ML) data models.
More recently, the company updated IBM AI OpenScale, a solution which tackles the problem of trust with a number of aspirational features. Those include understanding how an AI routine arrives at a given decision, identifying and rooting out bias in real-time, and ensuring that AI routines are fully auditable (key for regulatory concerns).
In order to bridge the skills gap within AI, IBM already offers a number of pre-built solutions, some specific to a given industry and others, such as Watson Discovery and Watson Assistant, which are more broadly applicable. But that’s nothing extraordinary. IBM’s true secret weapon behind its recent efforts to make IBM Watson more accessible doesn’t lie in technology. It rests in the hands of a small team of data professionals, the IBM Data Science Elite Team.
IBM offers this professional service offering as a set of free services ranging from 30-minute consultations to six-week on-premises engagements. There’s also a paid option for full 10+ week co-development efforts. For all options, the goal is to kick-start value of IBM Watson in a hurry.
And that’s the key. Project abandonment remains a huge problem for those seeking to make use of AI in practice. If IBM can turn even a few potentially lost projects around on its own dime, those customers and the industry as a whole will see that the value of AI is not found solely within software, but also within human expertise and insight.
IT Connection 