Microsoft Ignite: Harry Shum's keynote from the future

Microsoft EVP Harry Shum outlined the three big problems that Microsoft Research is thinking about today, so that 20 years from now, the solutions are inevitable, elegant -- and nearly invisible.

Lisa Schmeiser

May 4, 2015

4 Min Read
Microsoft Ignite: Harry Shum's keynote from the future
Microsoft press

Should anyone doubt the allure that Microsoft Research holds, consider this simple metric: EVP Harry Shum's keynote was scheduled for 5 p.m. on a Monday evening, and the room was packed to the rafters.

What's the big deal? Shum was providing a roadmap to what Microsoft's research division was devoting its energies to. He focused on two primary areas: an explanation for how Microsoft Research works to advance the three "big ambitions" articulated by Satya Nadella in the morning's keynote, and three big problems that Microsoft Research is tackling right now.

Those three big ambitions were to create more personal computing, to reinvent productivity within (and outside) the workplace and to build an intelligent cloud.

"It's really about getting more computing into everyone's daily lives," Shum said. And the key to moving toward a future of ubiquitous and unobtrusive computing? To accept that future breakthroughs are going to be practically invisible.

"We focus on solving a whole set of problems that most people don't even know about yet," Shum said.

But it's not just the problems that are nearly unseen. The solutions will be too, thanks to the one-two punch of the cloud and mobile computing.

Shum said, "Tomorrow's breakthroughs will be invisible for two good reasons: As computing shifts to the cloud … the kind of collective intelligence that we can harness from this huge amount of data, that is where we will see 'the magic/ happening. At the same time, we see that the user keeps moving, has entire computer ecosystem on the move."

It's also important to note that "tomorrow" does not necessarily mean "in the next 24 hours." Shum says that Microsoft Research is committed to the "long game," laying the foundational blocks for a new approach decades before there's any practical application.

And in the mind of Microsoft Research, the decade flies by. Shum paraphrased a sentiment he attributed to Bill Gates: People tend to overestimate what they can get done in a year but vastly underestimate what they can get done in 10 years. 

Shum then connected the dots from Microsoft's early research to its emerging applications for the consumer. He explained how some of Microsoft Edge's features -- like the ability to translate a web page from one language to another -- had their roots in natural-language work that Microsoft Research began 24 years ago.

From that past-is-prologue-to-the-product example, Shum then moved on to the three "big problems" that Microsoft Research wants to solve.

The first: The looming data-pocalypse. The sheer amount of data generated by smart devices is projected to reach north of 44 zettabytes, and being able to manage that data -- and make sense of any of it -- is a real challenge. One of the approaches being developed is data visualization: It's a way to make vast and complex pools of data make sense without too much work on the beholder's part. 

Here, Dave Brown tagged in to talk about NUIGraph, saying, "I'm fascinated by the potential of computer graphics to make sense of data." He perceives three ways graphics do this: by providing a way to sift data temporally, by using the invisible, intelligent cloud to sift through data sets and find patterns within them, and by providing a way to democratize access to data analysis beyond a small set of data scientists.

The second big problem that Microsoft's looking at is the end of Moore's Law. The division has both a short-term strategy and a longer-term play. In the shorter run, Microsoft Research is looking at programmable hardware; a prime example of this is their use of field-programmable arrays (FPGAs) in the servers that run Bing search

In the long term, the research arm is betting on quantum computing. How big a bet? In 2006, Microsoft's Station Q was founded to serve as a center for the study of topological quantum computation, or the notion of using a subatomic particle that had not yet been proven to exist as a means to do massively fast, massively scaled computations.

Microsoft Research is currently looking at quantum computers to "to develop entirely new ways of learning about data and thinking about ambient intelligence," said senior researcher Krysta Svore.

​And the final problem Microsoft research is tackling? The limits of human intelligence. Shum dismissed any concern that one day, we'd create AIs that overtook humanity -- a timely concern for anyone who spent the weekend at the cinema watching Avengers: Age of Ultron -- suggesting that instead, we'd work together and harness the rapidity and accuracy of software tools/AIs as tools for our own work.

The prime example of this in the keynote: computer vision, or training computers to correctly recognize and identify objects. At this point in the presentation, we saw a software tool that could identify individual elements in a photo down to the pixel -- demonstrating that computers are now more accurate at visual identification that people.

The applications of visual identification in conjunction with data management could lead to being able to calculate the calorie load in a given meal merely by snapping a photo, diagnosing patients via a selfie or identifying counterfeit drugs

The keynote ended with a tangible example of how Microsoft Research tackled the complex problem of being able to translate languages on the fly, showing two children speaking two different languages -- but holding one conversation on Skype. It was visible evidence of past Microsoft Research projects, and the end to an hour of pondering the invisibly changing future.

 

Read more about:

Microsoft
Sign up for the ITPro Today newsletter
Stay on top of the IT universe with commentary, news analysis, how-to's, and tips delivered to your inbox daily.

You May Also Like