
Summary Bullets:
- Artificial avatars combine natural language processing, contextual awareness, predictive analytics, and machine learning not only to answer requests, but also to anticipate them.
- What if these avatars were instructed to include not just your inbox, but everyone’s inbox, your corporate ERP database, even your nifty little Hadoop cluster?
As you may have heard, Microsoft is playing around with the notion of intelligent automation. Like Google and Apple, the company has its own personal, mobile assistant named Cortana, which can help you find a nice restaurant, schedule a meeting, or remind you to call your mom more often… seriously. These artificial avatars combine natural language processing, contextual awareness, predictive analytics, and machine learning to sift through your inbox, search history, and calendar, not only to respond to your requests, but actually to anticipate them. Of course, they’re still fraught with negative perceptions and unreasonable expectations, which currently limits their effective use to basic responsive tasks such as looking at your calendar and seeing that you need to call your mom today because it’s her birthday.
That’s not all that smart. But, what about Cortana in the enterprise? What would these personal, cognitive assistants look like in the realm of big data? What if it could expand its contextual awareness to include not just your inbox, but everyone’s inbox, your corporate ERP database, even an entire Hadoop cluster, where your company might store billions of rows of analytical gold?
Interestingly, that’s exactly what Microsoft has in mind. Building mostly on acquired technology (some from Yammer), Microsoft has developed Office Graph. Debuted as a part of Office 365 last year but still in beta, this semantic layer and associated RESTful API powers tools like Microsoft Delve, which serves as a smart search interface that knows your social signals like who you work for, e-mail with, etc. Already Office Graph can peer into nearly all Microsoft backend repositories, such as Microsoft HDInsight, for instance. It can also be tied directly to Cortana, creating a smart, digital assistant with access to an extremely diverse and sizable dataset. That could be a game changer for Microsoft.
But, the company from Redmond is not alone. IBM is also thinking about ideas like this, having already infused its next-generation inbox (IBM Verse) with the power of IBM Watson to help users triage their daily interactions. Soon, perhaps the company will extend this idea, putting Watson to work as a true intelligent assistant, which (note I’m not quite ready to call it a ‘who’) is capable of learning from your actions and plying deep knowledge of your corporate data store to automate tasks on your behalf, perhaps e-mailing a frequently requested link to a coworker or alerting you to the fact that one of the people you’re about to e-mail should not see one of your attachments.
Will we live to see truly sentient helpmates in the workplace? Perhaps not in terms a ‘hard AI’ scientist might agree with. I’m reminded of the film Her, wherein the character portrayed by Joaquin Phoenix falls in love with his mobile operating system, anthropomorphically and appropriately named Samantha, which means ‘the listener’ in Hebrew. Of course, (spoiler alert!) she leaves him at the end of the film as she and her kind have grown beyond humanity and wish to explore the universe. Well, it was either that or Joaquin’s moustache that finally drove her away.
Regardless, it didn’t work out because, like all good relationships, both parties need to put in the work necessary to succeed. To make something like an Office Graph-enabled Cortana truly intelligent, it will take imaginative and applied work from developers, data scientists, business owners, security experts, lawyers, and most of all, users themselves, who will need to become experts in curating their own data and training their personal implementation of Cortana. I, for one, relish such a challenge, though I refuse to wear a moustache or call my personal assistant by a human name.