Amazon’s Alexa is an ESTJ on the Myers-Briggs type indicator, according to Amazon VP of Alexa and devices Miriam Daniel. Those initials stand for extraversion, sensing, thinking, and judging. For anyone unfamiliar with Myers-Briggs, the test is said to be able to classify any person — or AI, in this case — according to 16 personality categories.
The Myers-Briggs Foundation website describes ESTJ personalities as practical and matter-of-fact, good at achieving the most efficient results possible and “forceful in implementing their plans.” Several U.S. presidents have been categorized as ESTJs, and this personality type is sometimes referred to as “the executive” or “the overseer.”
Daniel spoke Tuesday at VB Summit 2018, a two-day gathering of business executives to discuss the impact of AI. She described Alexa’s Myers-Briggs profile as more of a mental model, rather than a strict rule, for the personality team charged with making the assistant.
“We tried to say, ‘What would be a good representation of a personality?’ And we said, ‘Helpful, friendly, a little witty, has to be intelligent, has to be resourceful, humble’. So we kind of broke down some of the team at Amazon, as well as our customer base, and then said: ‘We’ve got to shape her personality around that’,” Daniel said in describing initial internal efforts to define Alexa’s personality. “That’s our north star, our long-term project. We think it’s going to be a multi-year effort to get her to be more like a friend or companion or colleague or somebody that you would really trust as your assistant.”
VentureBeat contacted Apple, Google, and Microsoft to ask if other tech giants think of their AI assistants as an INJF, ENTJ, or other type on the Myers-Briggs personality scale. We have not heard back regarding Siri or Google Assistant.
Cortana senior personality lead Chris O’Connor told VentureBeat in an email, “We tried to get Cortana to take the test, but she was so warm, empathetic, responsive, responsible, and highly attuned to the emotions, needs, and motivations of others that she didn’t have time.”
Personifying AI assistants
As silly as it may seem that Alexa can tell you her favorite color (ultraviolet) or sports team (Seattle Seahawks) or who she wants to win the Academy Award for best actress, these details are added to make her seem more human when she (Alexa is a female to Amazon, and a feminist) speaks with people. The team decides what Alexa says in response to personality questions, marriage proposals, and joke requests.
In pursuit of a more humanlike Alexa, Amazon has in recent months added a series of new features, like Whisper Mode and a follow-up mode so you don’t have to say “Alexa” after every exchange, resulting in potentially more conversational interactions. There’s also a feature called hunches that predicts your needs based on past activity.
This work is in addition to ongoing efforts to make Alexa better at understanding feelings and emotion from the human voice.
Each development, Daniel said, is an example of how Amazon is working to make Alexa respond in ways that are more like a human than a robot. If you whispered to a human that the baby is asleep, for example, they wouldn’t yell back at you at 100 percent volume.
Alongside people’s desire to control smart home devices with their voice, Daniel said user questions probing for personality clues was one of the biggest initial surprises for the Amazon team after the Echo first became available in late 2014.
People may justifiably presume Amazon built the Echo because CEO Jeff Bezos wants to sell more stuff. But Daniel said the goal was to create a minimal but lovable product and focus on the top 10 things people can do quickly, like listen to news, play music, or set timers, reminders, and alarms.
Though Alexa is an ESTJ, her answers to questions vary depending on where she is in the world. In the past year, Alexa speakers have arrived in France, Japan, and most recently Spain and Italy.
“[W]hat beer she likes in the U.S. cannot be the same answer in Germany, or what movie star she likes in the U.S. or any other country is not the same as the movie star she likes in India. If an Indian customer asks, you better give a Bollywood answer, and if you don’t say cricket, you’re done,” Daniel said.
The importance of personality for an AI assistant to create connections with users echoes statements made by others engineering conversational experiences for big tech companies, including eBay VP of applied research and machine learning Alexander Stojanovic, who joined Daniel on a panel about AI for product development at VB Summit.
During the panel, he discussed AI being used to detect fraud as part of a broader mission to earn trust.
“The difference between a toaster in the 1970s and, let’s say, an Alexa is the fact that there is a person, there is a personality, there’s an engagement and an emotional connection,” he said.
The future, he said, “belongs to those who build the deepest, richest emotional connections with customers.”
To Read Our Daily News Updates, Please visit Inventiva or Subscribe Our Newsletter & Push.