Skip to main content

Beyond the smartwatch: how invisible machines will shape Microsoft's future

Beyond the smartwatch: how invisible machines will shape Microsoft's future


A chat with Rick Rashid, chief research officer at Microsoft

Share this story

rick rashid microsoft research
rick rashid microsoft research

Rick Rashid has a lot on his plate. As the chief research officer at Microsoft Research, the division he founded in 1991, Rashid is charged with overseeing some of Microsoft's most daring and diverse projects. His vast network of researchers pioneered the motion-sensing technology behind the Kinect, and have since explored everything from augmented reality to wearable technologies. They’re even working on an interactive contact lens display that would be powered by a user's body heat.

But the company's most groundbreaking project may be one of its least flashy. Late last year, Rashid demonstrated a new technology that combines speech recognition with voice translation, converting a speaker's words into a different language in real time. During the demonstration, held at an event in China, Microsoft's software transcribed Rashid's words into Mandarin script almost instantly, before voicing them a few seconds later — in his own voice.

It was a remarkable presentation, and one that, according to Rashid, left several audience members in tears. In this instance, Microsoft's software translated at an error rate of seven percent (most human translators see error rates between two and four percent), though Rashid notes that there's still room for improvement. The company's "deep-neural-network" recognition still struggles to distinguish phrase breaks from sentence breaks, and has trouble discerning contextual meaning. Nevertheless, Rashid says his team has seen rapid progress in recent months, putting the vaunted language barrier well within striking range.

"I'm hoping that in the next few years this will be a solved problem," he said Tuesday during an interview at the Microsoft campus outside Paris.

Microsoft Research's speech recognition project is part of its broader efforts in machine learning — a field most commonly associated with artificial intelligence and other science fiction phenomena. The premise is conceptually straightforward: by programming computers to recognize historical patterns in vast amounts of data, he and his team hope to create machines capable of understanding — and evolving — in more human-like ways.

"We have a tendency to attribute anthropomorphic characteristics to things because that's who we are."

But as computers become more lifelike in their cognition, how long will it be before they become emotionally aware?

According to Rashid, his division has done extensive work in emotional recognition, developing algorithms capable of gauging a person's state based on the tenor of their voice, the speed at which they speak, or even the contours of their face. Microsoft and others have also explored ways to imbue machines with more emotional nuance, tinkering with different voices and intonations.

As we develop more meaningful relationships with technology, he says, it's only natural that we would begin to develop more emotional ties, as well.

"We have a tendency to attribute anthropomorphic characteristics to things because that's who we are," Rashid explains. "And there's a tendency to treat anything that's animated in the world as if it's something like us."

"It's much easier for me to predict where technology's going than what people are going to do with it."

From a corporate standpoint, it's not clear when any of these technologies will be deployed into the marketplace, though Microsoft Research may help the company carve a place in the wearable technology market. Thus far, the market has been spearheaded by companies like Fitbit, Jawbone, and Nike, with Google (and perhaps Apple) leading the development of new wearable form factors. Sources close to Microsoft have told The Verge that the Xbox team is currently prototyping its own smartwatch, though its timeline remains unclear.

Rashid didn't reveal any details on Microsoft’s roadmap during our interview Tuesday — "It's much easier for me to predict where technology's going than what people are going to do with it" — though he reiterated that his research branch has been involved in a range of wearables-related projects.

Microsoft Senior Researcher Desney Tan and scientists from the University of Washington have recently developed an electronic contact lens capable of measuring a user's blood sugar levels or other biometric indicators. Rashid tells The Verge that the team is looking to incorporate an electronic display into the lens as well, noting that it would be powered by a user's body heat, and that the technology could be modified for communication or other non-health-related purposes.

When it comes to smartwatches, Rashid strikes a more cautious tone.

"I think watches are certainly a possibility, although a lot of young people don't really wear watches," he explains. "One of the problems that we encountered when we were working on watches back in the early 2000s was that a lot of our hypothetical target market didn't actually wear a watch."

"It's more about data than it is about the device."

Ultimately, however, Rashid seems focused on a more distant future — one where consumers are constantly surrounded by interactive interfaces, and one where data transcends the confines of traditional tech hardware.

"Computing in the future is going to be about delivering a service to you through anything — any kind of device, whether it's your walls or laptop, car or tablet," Rashid explained. "And you won't really care about the physical characteristics of the device, per se. It will just be a transducer, an affordance, that gives you access to that."

"It’s more about the data and the service than it is about the device."