Why email is a better virtual assistant

1 google inbox

Google is inventing something better than Siri, Cortana, Alexa or even Google Now: artificial intelligence email.


Now that virtual assistants are here, we need to decide where they live. Here's an idea: How about inside email?

The location of virtual assistants depends on which assistant you're using. Siri, for example, lives on the other side of your iPhone or iPad Home button -- or the Siri button on the new Apple TV remote.

For Cortana, it's inside the Windows 10 search box on the task bar or conjured up with the microphone icon. You'll also find Cortana through a voice command on Android and, soon, living inside a dedicated iOS app. And if one report is accurate, Cortana will also live in your ear, accessible through a tiny hearable computing device.

Google Now can be found all over the place: in the Chrome browser, in the iOS Google app, inside Android apps via the Now on Tap feature, via a voice command on Android phones, on Google Glass and elsewhere.

And Amazon's virtual assistant, Alexa, can be accessed via the Amazon Echo appliance.

Interacting with a virtual assistant is mostly about talking. You talk to the assistant. The assistant (usually) talks back.

This would be super convenient if your assistant were available to you all the time. But it's not. If you're driving, or at the movies, or in a meeting or beyond a mobile data network, you might not be able to get access. Also: Talking is a natural user interface, but it's not easily reviewable, searchable or made asynchronous.

What if every virtual assistant interaction showed up in email? And what if your virtual assistant actually lived in email?

It appears that Google is exploring this idea.

Cheating on the Turing test

The Holy Grail for chatbots -- artificial intelligence software designed to hold conversations with people -- has been the Turing test, first outlined by British mathematician and computer scientist Alan Turing in 1950. A computer program passes the Turing test if, after a text-based chat conversation with a person, the human can't tell if he's interacting with a human or a machine.

No program has yet uncontroversially passed the Turing test. Natural human language is still too difficult a problem to solve for today's technology. And that might be a good thing for now. Once computers can pass as people, they'll replace people in a wide number of jobs at scale.

Meanwhile, it's possible to creatively apply today's not-quite-Turing-ready technology. Instead of replacing humans, what if chatbot A.I. technology was a tool for people to work faster, better or with more ease?

In other words, instead of stealing your job, what if A.I. helped you do your job better? Google is applying that idea with something called Smart Reply.

If you get an email that invites a concise reply, the mobile Inbox apps now suggest three possible replies for you. If you like one, just pick it and choose "Send."

This, by the way, is how human assistants work: "Bob Jones wants to move your meeting to next Tuesday. Should I accept or insist we keep the current time?"

Google implemented the Smart Reply feature last week for iOS and Android versions of its Inbox email app.

Here's how to understand Smart Reply: Think of it as a chatbot that "cheats" on the Turing test by getting a second opinion from a human, who can choose one of the three options or ignore them and write her own. Either way, these replies will convince the person on the other end of the conversation that the interaction is with a human, because it is with a human -- an A.I.-assisted human. Human judgment and self interest are the gatekeepers that choose a convincingly human reply.

Smart Reply currently has something like 20,000 possible responses and is designed to learn and become more naturally conversational over time. And it learns, by the way, without reading incoming emails. The system does keep track of the "triggers" directing how the reply options will be constructed and also which reply is chosen. By monitoring these relationships, Google can improve and optimize the responses continuously.

Responses are typically in the three-to-six-word range.

Specifically, Smart Reply uses two kinds of artificial neural network technology. One, called an encoder, understands the email you get, and the other, a decoder, builds the possible replies.

Smart Reply starts by deciding whether an email needs a reply or doesn't need one, and whether the email can be answered concisely. If so, it then zeros in on the meaning or substance of the email, rather than the syntax.

From the research blog post: "The encoding network consumes the words of the incoming email one at a time, and produces a vector (a list of numbers). This vector, which Geoff Hinton calls a 'thought vector,' captures the gist of what is being said without getting hung up on diction."

In other words, it's not an if-then rule-based system. It simulates understanding and responds accordingly.

A senior research scientist at Google named Greg Corrado told me that this "thought vector" is an abstract mathematical representation of intent. "The vector points in a direction in hundred-dimensional space, and messages with similar intents should have vectors which point in similar directions," he explains.

This is how the human mind works, by the way. No matter what creative, long-winded or poorly constructed request is expressed to us, we zero in on the substance of what is being asked and reply to the intent, not the form, and do so with a tone that makes sense based on the circumstances.

Google revealed in a blog post last week that an early version replied with "I love you" to most emails because it's such a common reply. That would have been exceedingly friendly, but probably not appropriate for most business communication.

Smart Reply's early inclination to express affection drove researchers to spend more time working on matching the tone of the reply with that of the initial email.

Why Inbox is a virtual assistant

"But wait," I can almost hear you saying, "semi-automated reply capabilities do not a virtual assistant make."

It's worth pointing out that Inbox already has assistant-like features, such as autocorrect-like suggested reminders. Inbox also populates reminders with publicly available information like phone numbers, directions or business hours when your reminder references a specific business.

Google also uses "natural language understanding technology" to identify incoming messages as "to do" items. So if your spouse, for example, asks you to do something, Inbox will add a one-click option in the subject view of the email to auto-create a reminder.

If you create a reminder in Google Keep, which is Google's service for holding arbitrary bits of information, Google Inbox will notify you when the user-specified time comes, with a link back to the Google Keep note.

A virtual assistant is a computer-based system that watches over you, protects you, warns you, reminds you and helps make your life easier and better. With Smart Reply, Inbox is heading confidently in that direction.

But unlike voice-command virtual assistants, such as Siri, Cortana, Alexa and Google Now, the virtual assistant in Google Inbox is integrated seamlessly into your regular, everyday communication. It offers the benefits of a voice-based virtual assistant, plus the benefits of email: It's asynchronous as well as easily reviewable and searchable.

In its quest to build a better email service, Google has also started to build a better virtual assistant.

If you're using Google Inbox already, it's not something you have to go find, download, install, learn or adapt to. It's just email with a virtual assistant that uses artificial intelligence to help you out.

This story, "Why email is a better virtual assistant" was originally published by Computerworld.

ITWorld DealPost: The best in tech deals and discounts.