Microsoft and Google laid out their respective visions for the next generation of virtual assistants this week, with enhancements to Cortana and Assistant, respectively, designed to support workplace productivity.
Though voice assistants have yet to arrive in force in the workplace, analysts see plenty of potential. By 2021, 25% of digital workers are expected to use a virtual assistant on a daily basis, according to analyst firm Garner, up from 2% who do so now.
On Monday at its Build developer conference, Microsoft offered a glimpse of where Cortana is headed, with plans for the virtual assistant to support more fluid, back-and-forth conversations.
A demo video during Microsoft CEO Satya Nadella's keynote presentation showed an office worker managing a calendar via Cortana on a mobile device. In the demo, Cortana answers numerous questions and carries out multiple actions during a long, continuous interaction with the user that lasted more than 30 “turns.” That included viewing upcoming plans, scheduling and rescheduling meetings, checking availability and booking rooms. The user is also able to to pull in relevant information such as local weather and traffic information, and send directions to an in-car virtual assistant.
Although Cortana can already carry out many of these actions, they tend to be done separately. Dialogue does not flow in a natural way.
“In spite of all the progress [with virtual assistants] you have to remember that most of the conversations we have today are still very brittle. They are not multi-turn. The context from turn to turn gets lost,” Nadella said during his keynote.
“Human language is complex, the context is subtle,” he said. “So how do [we] make sure that the natural language capabilities inside these personal assistants is capable of having that shared context across long dialogue, versus just a few ‘turns’?”
The ability to engage in more complex and dynamic interactions improves on current iterations of AI assistants by keeping track of context, said Nadella. That advances follows from Microsoft’s purchase last year of Semantic Machines, a conversational AI startup.
Appealing to business users
Once the technology is integrated into Microsoft products, users will be able to retrieve information from a variety of applications, with Cortana acting as the interface.
“We want it to be less cognitive load, less feeling like, ‘I have to go to PowerPoint for this or Word for that, or Outlook for this and Teams for that,’ and more about personal preferences and intents,” Andrew Shuman, Microsoft’s corporate vice president for Cortana, said in a blog post.
Microsoft is confident of Cortana’s future in the office. Although it’s available on 800 million Windows 10 devices, it has largely failed to capture consumer mindshare as much as rivals. There are now plans to connect Cortana with other AI assistants rather than compete head-on; Microsoft struck a partnership with Amazon to integrate Cortana with Alexa in 2017.
Patrick Moorhead, founder and president of Moor Insights & Strategy, said Microsoft has been wise to focus its attention on office duties rather than chasing the consumer market. “I believe the business focus is the right move as this is what the company can win here,” he said. “…I would question the company’s chances in consumer.
“Microsoft has many of the most popular mobile apps with Office 365 and I believe if businesses offer a super-tuned Cortana for their work, workers will use it.”
JP Gownder, vice president and principal analyst at Forrester, said the Cortana updates can help Microsoft convince workers to use its virtual assistant more regularly as an interface with its applications.
“I think the new Cortana functionality makes sense – making calendaring smarter, for example,” he said.
“The new Cortana almost feels like a feature of Office rather than a stand-alone virtual assistant, which isn't necessarily a bad thing,” Gownder said. “Integrating Cortana's intelligence and natural language capabilities into common business applications like Office would save employees time and drive productivity improvements.”
He noted Microsoft still needs to show it can deliver on the promise of more natural conversations showcased in the Cortana demo. “Conversational intelligence is far too challenging for us to take their word for it.”
Google unveils smarter, faster Assistant
Microsoft is not the only tech firm hoping to popularize virtual assistants in the workplace.
Amazon has Alexa for Business as its beachhead into the enterprise, offering management tools to support deployments of thousands of Echo devices into offices and meeting rooms.
Apple, whose strategy relies on sales of HomePod devices, has not shown a strong intention to target Siri at the workplace – yet. It did, however, announce a partnership with Salesforce last year that will see the CRM vendor integrate Siri deeper into its mobile apps, which are often used by sales and marketing professionals. And Apple’s reputation for stringent protection of customer data could help allay privacy worries about deploying voice-activated assistants in the office.
Meanwhile, Google Assistant, typically seen as the strongest competitor to Alexa in the consumer arena, has indicated plans to tailor its virtual assistant for business use. That lines up with plans, announced at Google Cloud Next last month, to link Assistant with calendars in G Suite, Google’s business productivity app suite.
At this week’s Google I/O developer conference, the company talked up the “next generation” of Google Assistant, which is faster and more personal.
By running machine learning models locally on a user device, such as Google’s Pixel smartphones (where it will appear later this year), the new Google Assistant should deliver results up to 10 times quicker than via Google’s cloud servers. That means processing speech with close to zero latency, allowing transcriptions to occur in real time, for example.
This has been made possible by “advances in deep learning” that enables the machine learning models required for speech recognition and language understanding to be condensed from 100GB to less than half a gigabyte, Google CEO Sundar Pichai said during his keynote presentation.
“It makes the assistant so fast that tapping to use your phone would seem slow. I think this is going to transform the future of the Assistant,” Pichai said.
That lets users quickly switch from different tasks, such as checking the weather, playing music, ordering taxis, setting timers, as well as opening, navigating and sharing data between apps instantly.
The ability to run some or all of the data processing on devices is important for three reasons: faster response time; improved security and privacy – as sensitive data doesn’t leave the device; and less dependency on connectivity bandwidth, said Annette Jump, senior director analyst at Gartner.
And in an office environment? "In the workplace, it can certainly help with storing sensitive data around passwords, voice authentication, etc., locally,” she said. “In the next few years, Google Assistant can also control [a] broader range of smart devices in the office, like lights, conference room devices, heating, smart screens/boards.”
A demo during the I/O keynote also showed how Google Assistant can draft and send emails using only voice commands, which requires the assistant to discern when a user is selecting a subject title, writing a greeting and moving to the next line of text.
The company also offered updates to Google’s Duplex technology, showcased at last year’s I/O, which can make restaurant reservations on behalf of a user. Duplex will soon be extended to the web, the company said Tuesday. Initial use cases include automated booking of online car rentals, with information – including dates of the trip and even type of car – and entered on behalf of the user, who can then confirm the details before finalizing. Those actions are carried out via Google Assistant.
As virtual assistants become more commonplace, both in work and home environments, it's likely Google Assistant will become more entwined with work apps.
“Voice interactions are very natural to humans and as we start using more VPA speakers in our homes, we would find the desire to use voice as a way of interaction with devices in the workplace, too (like with touch interfaces),” said Jump.
“...I would expect Assistant to be integrated in more productivity apps and devices from Google and be used by both mainstream office workers, as well as front-line workers or remote workers using new type of devices, like smart glasses or VR-headsets.”