Credit: Florence Ion
When users first try virtual assistants (like Siri, Google Now, Cortana, M or Alexa), they're struck by the idea that they're talking to a computer, rather than a person.
That much sounds obvious. In reality, however, the responses from virtual assistants are far more human than most people assume. In fact, every response is carefully crafted by a person or a team of people.
What you get as a response to your question or request to a virtual assistant isn't what a real-live person said. It's what a team of people believe a real-live human being could or should say.
Some replies are constructed from prerecorded words and phrases -- the sentences are pieced together by software to answer some arbitrary question -- and others are written as full sentences or paragraphs. Let’s take a look behind the scenes and see how this works.
Speaking at the recent RE-WORK Virtual Assistant summit, Deborah Harrison, a writer on Microsoft's Cortana team, revealed that in the U.S., there are eight full-time writers chugging away every day on choosing what Cortana will say. These writers are novelists, screenwriters, playwrights and essayists. According to Harrison, they start each morning with a meeting that lasts an hour or so. In that meeting, they pore over information about what people are asking or saying to Cortana, looking for problems that can be solved with better replies. (Yes: All your interactions with virtual assistants are recorded, cataloged and studied). Then the writers spend the rest of their day crafting the best responses.
In fact, the writing done for Cortana (and Siri, Alexa and other virtual assistants that have personalities) is a lot like novel writing. There's no story or plot, but there is one character, which has a gender, an age, an educational level, a particular sense of humor and more. That human-created character or personality has to remain consistent across queries and contexts. When the writing staff crafts canned replies, it's essentially dialogue based on the character they've developed.
For example, the Cortana team will take the simplest reply and agonize over whether the answer is meaningful and relatable for people from a wide variety of cultures and backgrounds and age groups. It can't be sexist or even encourage sexism. Nor can it be militantly feminist, overtly political or even gender neutral (Cortana is not human, but she is female, according to Harrison).
Cortana even holds what some might consider religious or political views. For example, Cortana supports gay marriage.
There's always a balance to be achieved. And having a good sense of humor is important. Cortana must be funny, but sarcastic humor is off the table.
Sign up for CIO Asia eNewsletters.