Microsoft's 'AI for Everyone' Plans Detailed at Live! 360
A Microsoft program manager this week gave some insight into the ways that artificial intelligence (AI) capabilities are shaping the Microsoft stack -- sometimes in surprising ways.
Pranav Rastogi, who led Tuesday's keynote of the inaugural Artificial Intelligence Live! track at the Live! 360 conference, is one of the people inside Microsoft helping drive those capabilities and technologies across the company's vast array of products. Rastogi provided attendees with an overview of what those technologies are and where they're starting to emerge in products.
"The idea here is really to democratize AI for each and every employee so that it's available and employees can use it to transform their own businesses," Rastogi told an audience of several hundred attendees at the Orlando conference.
During his hour-long talk, Rastogi provided a tour of AI technologies that can immediately be leveraged by developers, end user and business analysts. To date, AI has mostly been the domain of data scientists. Rastogi's discussion dealt with the other user profiles who may not think of themselves as potential users of AI right now. An example was a slide labeled, "Introducing the Citizen Data Scientist."
All of the AI technologies he highlighted fit into a bucket that Data Relish Ltd. Principal Jen Stirrup, another speaker at the conference, described Tuesday as the types of machine learning capabilities that are commonly coming online right now -- training computers to do a single task at roughly a human level of proficiency. That's as opposed to the strong AI of self-directed fictional scenarios like R2-D2 in "Star Wars," Skynet in "The Terminator" or HAL 9000 in "2001."
For Microsoft, the AI democratization journey has three phases. First is infusing every Microsoft application with some AI capabilities so that early adopter customers can leverage the technologies if they're looking for them. The second phase involves bringing AI to every business process, which would mean driving adoption among users both through increased ease of use and raising awareness of the vertical and horizontal benefits of using Microsoft's tools. The final phase is getting every employee at all of Microsoft's customers using the AI capabilities in some way.
The "every application" phase is in the early stages but spreading quickly across many products, making the effort already broad, if not particularly deep. As an example, Rastogi showed how Microsoft is redefining existing applications with AI using the pre-built AI services, such as Vision, Speech, Language and Search. Those capabilities are being used to create new conversational experiences inside other applications like Microsoft's own Cortana, Office and Skype, as well as other applications like Slack, Facebook Messenger and Kik Messenger.
Rastogi also showed how dense the company's flagship AI platform, Azure, is getting with machine learning capabilities. At the first level are the sophisticated pre-trained models that are ready to be called from within other applications, such as the Vision, Speech, Language and Search services mentioned earlier. The lengthy list of Azure services also includes a few designed to help data science and development teams, such as Azure DataBricks, Azure Machine Learning and Machine Learning VMs. Additionally, Rastogi highlighted the Azure options for using AI-optimized hardware in Microsoft's datacenters, and for having the compute performed in the cloud, on-premises or at the edge.
The product set where the "AI everywhere" story appears strongest is in Power BI, Microsoft's business intelligence platform for accessing, manipulating and visualizing data. A product that essentially aimed to democratize BI is now evolving to do the same for AI, as well. There are capabilities for data scientists, certainly, including Power Query integration for Azure Machine Learning and integrations with Azure frameworks. Data scientists and BI professionals can also script in R or Python or create machine learning models via clicking.
But end users also have ways to explore AI through Power BI, using Natural Language exploration. Examples of the types of things that end users or business analysts can leverage in Power BI include sentiment analysis, key-phrase extraction, optical character recognition and text translation.
Most of the AI capabilities Microsoft enables today still require a lot of leading-edge expertise, integration, development work and data science expertise. Yet it's clear that Microsoft is working rapidly to integrate those technologies all the way out to end-user-facing applications and will continue to push hard in that direction.
Posted by Scott Bekker on December 05, 2018