This article is part of Fast Company’s editorial series The New Rules of AI. More than 60 years into the era of artificial intelligence, the world’s largest technology companies are just beginning to crack open what’s possible with AI—and grapple with how it might change our future. Click here to read all the stories in the series.
“It’s better than Wite-Out.”
That’s not the way you’d expect billionaire and mogul Bill Gates to talk about the current state of business software—especially since Microsoft’s productivity tools, including Word and Excel, helped build his fortune.
But Gates believes that today’s business software isn’t much better than business tools from pre-digital times. “I’d say most of the opportunity to make computers improve work is in front of us rather than behind us,” he told a room full of Microsoft researchers and academics at the Microsoft Research Summit in July.
While Microsoft Office is definitely way more powerful than Wite-Out, Gates is right to be optimistic about the ways that computers can continue to change the way people work. Artificial intelligence could soon bring productivity tools to an inflection point. After years of slavishly digitizing and formatting words, numbers, and images, future versions of Word, Excel, and PowerPoint will be much more aware of your work and how you do it. Intelligent algorithms will find patterns and meanings in the data, and use the insights to help you get through your day more efficiently.
This transition to smarter work software has already begun. Microsoft’s Cortana personal assistant can scan your email and remind you of commitments you’ve made. The MyAnalytics tool in the company’s enterprise software Microsoft 365 knows all about how you’re spending your time. A new PowerPoint feature will advise you on the pacing of your presentation.
But Microsoft researchers are looking beyond specific features to help today’s knowledge workers. They’re also striving to understand the nature of work itself in the 21st century. The company is looking to AI to help break down work into smaller tasks that fit in the slots of available time during our increasingly fragmented workdays. Microsoft wants to build tools to help people squeeze more out of their time in an age of distraction—a strategy that may be the key to a new, more intelligent generation of Office.
Software in a time of information overload
Workaday business software has already changed a lot in the internet age, and will change further with the advance of AI, heavy use of contract labor, and the gig economy. Mobile devices let us get more work done away from a desk, but they also mix our personal lives with our work lives on one device.
With the rise of mobile phones and working across devices, the level of distraction coming from technology has reached a fever pitch, both in the office and outside. In a survey of more than 500 workers, RescueTime found that only 10% of respondents felt like they were in control of how they spent their time at work. Research from University of California Irvine professor Gloria Mark shows that people switch computer screens an average of 566 times a day. They check email 77 times a day. Facebook users in her sample checked the social media site an average of 38 times a day.
It’s not just the information overload of the internet that’s driving attention spans down.
“In the information workplace, people are constantly juggling competing demands, they’re reprioritizing tasks,” Mark said during a presentation at Microsoft’s Research Summit. “Research has shown that the scope of work has expanded, so people are actually working on more, different projects that they switch to in the workday.”
That means that work is faster-paced, more information-rich, and more varied than ever before. We’re doing a lot of work, more kinds of it, and we’re doing it in smaller chunks.
And yet Mark’s research shows that people seem to measure their own productivity by an older metric: how long they are able to give all their attention to a single task. Meanwhile, they struggle to block out time on their calendars for this kind of head-down work.
The days of focused, deep work may be over for many of us. Microsoft Research chief scientist Jaime Teevan believes we should stop trying to fit old workflows into the new fragmented reality of the workday.
“We think so much about people being in the ‘flow,’ but really, it’s hard to do big tasks; it’s hard to get into the flow,” says Teevan. “We have, on the other hand, lots of little bits of time, and we can make our tasks really small so that they fit with our small bits of time.”
Teevan has been publishing research on microproductivity and how it can be applied to tech products since 2014. “We’re thinking very much about how we can specifically help people get through their tasks with an awareness of . . . how they think about tasks, and in particular with an awareness that our lives are fragmented,” she says.
Microproductivity in theory
Microproductivity means dividing up work into a series of small tasks (“microtasks”) that require little time to complete, and build toward the completion of a larger goal.
We already do this, if not in a systematized way—like when we use the five-minute walk between buildings to answer an email. Teevan suggests more of our work can be broken into these kind of microtasks.
Teevan’s research has shown that performing microtasks can be a good way to ease into more demanding tasks that take longer. The research also shows you can benefit by starting work on a large project by completing some of its less-demanding microtasks, then moving to more demanding ones as you settle into the project.
This is also where AI can start to play a role. Some aspects of larger projects could get carved off as microtasks and automated, Teevan says, leaving the humans with the more creative and engaging work.
However, Microsoft researcher Shamsi Iqbal suggests that microproductivity might not define everyone’s whole workday. Some people’s days might be a mix of highly fragmented tasks and extended periods of work on a single task. Microsoft already has features that help keep people engaged in larger tasks. Windows 10’s Focus Assist feature, for example, can be configured to block various kinds of notifications and alerts when you need to stay in the flow.
If microproductivity sounds to you like a corporate plot to fit work tasks into every minute of every day, you’re not alone—it’s a common reaction to the concept. But the researchers say microproductivity really isn’t about more work. It’s about getting the same amount of work done more efficiently. The well-proven need for free time and family time doesn’t change. Moreover, the same microproductivity approach can be used for personal life tasks, like organizing a vacation or planning a dinner, says University of Michigan researcher Walter Lasecki.
Instead of stealing away free time, microproductivity is more about giving people control of how they spend their time at work. “You’re giving people the tools they need to structure their own workflows,” he argues. “Letting people have control over how, when, and what their work looks like is what’s exciting to me.”
The data-driven workday
Microsoft is just getting started applying its considerable AI know-how to its productivity tools. You’ll increasingly see the company introducing new features driven by what its algorithms have learned about your preferences and workflows, and about work itself.
Already, it’s gathering some of this data and organizing it via a tool called MyAnalytics in Microsoft 365, which tells you the amount of time you spent in “focused” mode, collaborating with others, and doing something other than work. It gleans these figures by scanning your emails, meetings, calls, and chats. It can also remind you if you have an outstanding request from a coworker, or warn you when you’re about to send an email to another person after-hours and suggest letting it wait until the morning.
Many features are driven mainly by data collected from your email—provided you opt in. Cortana can look through your inbox for commitments you’ve made and remind you to fulfill them. Outlook’s Focused Inbox helps you prioritize your email by identifying potentially important emails from people you collaborate with a lot, and by identifying newsletters and machine-generated mail and moving them down in the stack.
Microsoft researchers such as Teevan and their partners in academia are also thinking about how machine learning can help facilitate microproductivity. Lasecki said AI might analyze the work you’re doing and then work with you to break it into a to-do list of smaller tasks. If you’re a programmer building a new website feature, the project might logically break down it into subtasks such as designing the interface, completing documentation, pulling down open source code, and doing research. Based on its knowledge of those tasks and your preferences, work habits, and schedule, the AI could also help you figure out the right tasks to tackle at the right times. It might understand, for example, that your most productive times of day are midmorning and late afternoon, and suggest working on the tasks that require the most creativity and concentration at those times.
This ambitious form of AI hasn’t infiltrated your Office tools quite yet. Microsoft wants to be thoughtful about how it translates that research into real products. It’ll likely happen gradually, though Teevan says Microsoft is trying to shorten the process of bringing research ideas into core apps such as Word, Excel, PowerPoint, and Outlook.
To do this, Microsoft is asking the research community for help applying its microproductivity research to actual technology. It’s offering $1 million in funding to researchers developing tech that uses machine learning and human input to divide large projects into microtasks. It is hoping to fund tools that figure out which tasks are easier to do on mobile devices, or ones that can be started on one device and finished on another.
Though these research aims are about going beyond today’s business software, they aren’t purely speculative. “The research that we all are doing on the future of productivity, it’s not a thing about the actual future,” Teevan said in a recent podcast. “It’s directly relevant to the products we are building now.”