Forget AI and chatbots! Over the last year or two, a new business tech trend has emerged: voice. From Alexa to Siri, people everywhere are embracing smart assistants. In fact, 47.3 million adults in the US now have access to a smart speaker. That’s 20% of the country’s adult population!
Worldwide, 1.8 billion people are projected to use a smart assistant by 2021. As adoption continues to grow, there’s been a lot of speculation about how these digital helpers will change the way we work. EVP of AI and Research at Microsoft Harry Shum, for example, believes that in 20 years digital assistants will be so tightly intertwined with our lives, they’ll be like our alter egos.
In the future of work, will smart assistants ever do more than help us manage our calendars and order things from Amazon? What’s their potential for use within the enterprise? And are there any potential downsides?
Related: What the future of automation looks like
Accuracy (and Options) Are Getting Better…
A big question for most prospective users is: can smart assistants accurately provide the information you need? The answer is “sort of.” In 2018, for example, Siri is only able to answer about 40% of general knowledge queries. Of those questions answered, it gets roughly 80% correct.
Compared to last year—when Siri could only answer 31% of questions it was asked—its available knowledge base has grown remarkably. But accuracy has actually decreased; in 2017, even though Siri could answer fewer questions, it got a higher percentage (86%) correct.
Siri was the first smart assistant to go mainstream, and it still dominates the market. But today, almost every major tech brand offers a smart assistant. The biggest players are Amazon’s Alexa, Google Assistant, and Microsoft’s Cortana.
Users have much more choice today than they did three years ago—which means that you can choose a smart assistant that best fits your needs. Accuracy is much higher for Google Assistant, for example, which is natively integrated with their world-class search engine.
…But AI Still Needs to Improve…
Ever wonder why sometimes, your smart assistant seems to be less-than-perfect? Maybe you asked Siri for a vegetarian place to eat lunch with your team—and it suggested a steakhouse. Or perhaps you had to repeat yourself five times before it understood what you were asking for.
That’s because the technology behind these digital helpers is complex—and far from perfect.
Smart assistants rely heavily on artificial intelligence (AI) to create personalized, useful experiences. AI is the force behind Natural Language Processing (NLP), which allows your assistant to understand human language the way it’s actually spoken. It’s also what allows your assistant to learn from its experiences with you and get smarter over time.
By 2021, 1.8 billion people will use a smart assistant.
Despite all the hype surrounding it, AI still has a long way to go in terms of being a substitute for human interaction. It’s not great at soft skills, can be difficult to train, and often has trouble with complex statements and reading between the lines—in short, all the skills that are so highly prized in human assistants. (Case in point: veteran journalist John H. Richardson recently wrote about how frustrating an AI-powered scheduling tool was to use.)
That’s something that the companies who build smart assistants hope to change—and soon.
“We see this space as more than just building a personal assistant feature,” writes Marcus Ash, a Group Program Manager who works on Microsoft’s Cortana. “This is a long-term opportunity to build the first contextual operating service: a system that knows you so well that it can predict your patterns to help you get things done.”
…and Automation Is a Necessity, Not a Luxury
From a business perspective, however, smart assistants still have a way to go before they can have real value. Beyond checking your calendar or sending quick email replies, most smart assistants can perform only a limited range of business tasks. They’re just not super useful (yet) when it comes to business-critical processes, like searching for leads in Salesforce or creating invoices in NetSuite.
In order to become widely used in a business context, these tools need to not only integrate with enterprise apps—they also need to be capable of executing workflows that span those apps.
Some smart assistants, like Amazon’s Alexa, offer developer kits that allow you to build your own integrations and “skills” for your assistant. But that’s a technical challenge that few non-developers feel comfortable tackling, which means it’s not a viable option for many businesses who have limited dev resources.
A few middleware platforms—such as IFTTT—also offer pre-built connectors for smart assistants, but these often only work with a limited range of apps and may not feature enterprise-grade necessities like adequate security.
But there’s reason to believe that tech companies see enterprise integrations and automations as the next frontier in smart assistant development.
“An interesting front-runner in closing this gap is Microsoft,” says James Haight, an analyst at Blue Hill Research.
“Microsoft has made huge strides in its AI efforts with its Project Adam—and it is continually tying it together with its Cortana technology as a front end. We can already do simple things like talk to an Xbox, but the dominance of Microsoft’s office tools like Word, SharePoint, and Outlook point to a much larger impact.”
The Future of Work Takes Time
Smart assistants are just like any other tech development: they’re cool and have the potential to be really useful, but they’re also nascent. Currently, they’re improving at performing low-level everyday tasks. Though they still can’t pick up your drycleaning, they can help you save time, track your calendar, and find something good to eat for lunch.
Over time, their business potential will likely improve as well. As adoption continues to grow and they become an indispensable part of our lives, businesses will have more to gain from implementing them. That may spur on the expansion of their integration and automation capabilities—which, from the looks of it, is already in motion.
