A little more human

Siri is dying. Long live Susan Bennett.

16 min read

She’s been in the hands of over 100 million people. Perhaps she’s slept on your nightstand. She may have even drunk-dialed your ex.

And guess what: Susan Bennett, the original voice of Apple’s Siri, never saw it coming.

How do you become the most popular voice of the most successful tech company in the world—without your knowledge?

I interviewed Susan about her stumble into fame, then plotted her life against the rise of more human-like virtual assistants. It turns out that her voice appeared at two pivotal moments in the development of talking tech.

It makes you wonder, was Susan destined to become the world’s most recognizable digital voice?

Let’s start at the beginning.

A voice for an analog world

Susan was born into an analog world, where transistor radios carried the tunes of Elvis and Chuck Berry, and dialing a zero on a phone meant tracing a large circle around a rotary dial.

What was Susan up to?

But she did. In the late 1960s and early ’70s, Susan toured as a backup singer for Burt Bacharach and Roy Orbison. She also sang jingles on radio and TV commercials.

Then one day, a voice actor didn’t show up at the recording studio. Susan got pulled in to substitute. And she was good.

Accident number 1: a voice acting career.

And over in a parallel universe, visionaries and engineers were starting to shape Susan’s digital future.

Dreams of intelligent machines

Just after Susan was born, Alan Turing asked his famous question: “Can machines think?” He answered with the Turing Test: the ultimate benchmark for computer intelligence is the human-like use of natural language.

It was an ambitious vision at a time when you talked to computers through punch cards and blinking lights. But in the 1950s and ’60s, research centers like MIT, IBM, and SRI International were giving human-computer interactions a facelift.

Leading the way at SRI was Douglas Engelbart, the godfather of user interface design. In 1968, he brought together the first computer mouse, keyboard, hypertext, and videoconferencing into the “the mother of all demos.”

This motivated Xerox PARC to create the graphical user interface (GUI), which Steve Jobs mythically stole for the original Macintosh.

SRI would later begin another project to create the first intelligent, digital assistant. And they’d spin off two companies—Nuance Communications and Siri, Inc—that would come together to give Siri her brain and voice.

Meanwhile, Susan made her debut as a machine.

Tillie the first machine personality

In the mid-70s, a new kind of computer started appearing on street corners: money machines.

The utility of the machine was clear: money on-demand. But people were used to human bank tellers, and ATMs just didn’t feel right. Enter Susan’s voice.

Accident number 2: Susan’s life as a machine personality begins.

Back to our parallel universe, where a young Steve Jobs was beginning his quest to find the ultimate user experience.

Siri’s progeny

Around the time Tillie hit the streets, Steve jobs was soul-searching in India. When he returned, the first personal computer—the Altair 8800—hit the market. Soon after, Jobs cofounded Apple Computer with Steve Wozniak to get in on the action.

His first revolution came in 1984 with the original Macintosh. Jobs married SRI’s mouse and Xerox PARC’s graphical user interface into an iconic machine. Farewell DOS commands.

In 1987, Apple outlined the future of intelligent virtual assistants with the “Knowledge Navigator.” It organized your calendar, read your emails aloud, and hosted video conferencing. And it felt human—complete with a face, a voice, and personality.

But there was a huge gap between that dream and the technology of the time. Getting there required major leaps in voice recognition, natural language processing, and machine learning algorithms.

And of course, just the right voice.

“Beautiful woman who guides you to victory”

In the 1980s and ’90s, Susan’s voice began appearing on GPS and interactive voice response systems (IVRs). You know them as those painful automated phone systems for checking bank balances and flight reservations.

Then in July 2005, Susan started recording a new project that would change her life. The job? Nearly an entire month repeating nonsense phrases like:

No, Susan wasn’t recording the audio for some psychedelic Tim Burton film.

That text-to-speech company was ScanSoft. And–spoiler alert—the phone messaging system would end up being Siri.

But Susan had no idea. To her, it was all just gibberish, “the complete opposite of creative,” as she put it. So Susan took her hourly wage, and shot down a 5-year contract to continue working with the client.

But her “shrodding” and “shreeding” had already provided ScanSoft with all the sounds of the English language. Enter the ubergeeks, armed with a new type of speech synthesis called concatenation.

So, did Susan actually record all of Siri’s sassy responses?

And those programmers had just picked up $200 million to create a highly intelligent, personable, virtual assistant for smartphone consumers.

Where Siri got her smarts… and tongue

Remember SRI International? In 2003, SRI picked up a 5-year DARPA grant to build out project CALO—the “Cognitive Assistant that Learns and Organizes.” It was the Manhattan Project of artificial intelligence.

Leading the 300-person team was AI pioneer Adam Cheyer. Here’s how he described the project:

In 2007, Cheyer left SRI to form Siri, Inc, along with Dag Kittlaus and Tom Gruber. Soon after, Cheyer’s phone rang:

“Hey, it’s Steve. What are you doing tomorrow? Want to come over to my house?”

In 2010, Apple bought Siri, Inc, and Cheyer moved in as the Director of Engineering for the iPhone/iOS team.

Jobs also knew that Siri’s AI needed a brilliant voice interface.

Enter voice recognition giant Nuance Communications. Nuance had just merged with ScanSoft, who was holding a month’s worth of Susan Bennett’s sounds.

Apple’s Knowledge Navigator now had its brain and its voice. And Susan was in for the surprise of her life.

Siri, where did you get your name?

“This sounds like a person in my phone”

In the six years after ScanSoft studio boredom, Susan sold her voice to Coke, IBM, Ford, and other giants. She also guided millions of passengers a year as the voice of Delta Air Lines. Not bad for an invisible voice.

And then it came—October 4, 2011—the day speech joined touch as a viable way to interact with technology. Siri was famous, and Susan was her voice.

How did Susan feel about this? “It was a little creepy,” she told CNN in her coming-out interview.

But Siri was different.

Women’s health, sex, and drugs—Siri had answers for all those things. And many were answers that Susan herself would never give.

But Susan stayed quiet. For two years, she held her feelings inside while the rest of the world fell in love with her.

Why Siri was so special

Siri was far more than a novel voice interface. People connected with her.

Just ask Susan. She’s received thousands of calls and emails from people talking about the emotional relationship they’ve formed with Siri. Here’s why:

1
A voice that sounds human

I’ll let Susan explain this one:

But a beautiful voice doesn’t make an idiot attractive. Siri’s real charm is in her verbal behavior.

2
Social intelligence and contextual cues

Remember Clippy, Microsoft’s annoying virtual paperclip? You said go away, and he came right back. Microsoft killed Clippy in 2007. Lesson learned.

Siri embodies social cues that Clippy lacked. Like asking a follow-up question if she doesn’t understand, but without imposing herself on the user.

3
Personality and humor

Everyone has their own personality, and virtual assistants need one too. So Apple made Siri likeable by endowing her with her own sassy style. Some of my favorites:

Q: I love you Siri. A: I bet you say that to all the Apple products.Q: Siri, what do you look like? A: To tell the truth, I'm rather abstract-lookingQ: Siri, tell me a joke? A: The past, present, and future walk into a bar. It was tense.Q: Siri, what's the meaning of life? A: I can't answer that right now, but let me write a very long play in which nothing happens.

 

4
Learning from experience

The more you talk to Siri, the better she understands you. Isn’t that what friends do? Sure, sometimes Siri screws up. But you can tell her this, and she’ll remember it forever.

5
Managing expectations

People automatically apply the rules of social interactions to any engagement that feels human. So Siri avoids mistakes that humans would never make, like repeating identical responses when asked the same question.

Educator, angel investor, and author of Hooked, Nir Eyal, calls this “variable reward,” an essential ingredient in habit-forming products.

No, Siri didn’t do any of these things amazingly well by human standards. But she was lightyears ahead of anything else on the market.

But soon, her limitations became apparent.

Simpsons siri

An exit to new beginnings

Apple never admitted that Susan was Siri’s voice, they never paid her beyond the hourly wage in the recording booth, and then–they dropped her. From the iOS7 onwards, all of the original Siri voices were replaced on new iPhones.

Then a 2013 Verge video got everybody wondering about the human behind Siri. So with some prodding from her family, Susan made a decision that threw her quiet life into the spotlight.

Finally, two years after finding out she was the voice of Siri, Susan spoke up. A tough decision, but the move opened up a whole new dimension to her voice career. Like introducing Steve Wozniak at the 2013 Dallas Digital Summit, and presenting her story on the TEDx stage in 2016.

So what does Susan want now? Time to play music with her husband and their band. And definitely a chance to show off some of her other non-robotic voices. She’s got quite the repertoire:

Susan’s also studying Spanish, y habla muy bien. Si, Susan is so much more than Siri.

And she’s also reconciled her relationship with Apple, as she jokes:

Now the big question is, who will be leading the next-generation of smart assistants?

Is Siri dying? The future of intelligent assistants

So what’s next? A big global brain that solves problems on the fly. That’s why Dag Kittlaus, Adam Cheyer, a third of the Siri team left Apple soon after Siri’s launch.

On May 9th, 2016, Siri’s creators unveiled the future of intelligent assistants: Viv. Think of it as the new interface to the Internet—where all the apps you now interact with come together.

Cheyer describes it as part of a Browse-Search-Solve paradigm shift:

> Browse: humans do all the work by scrolling through posts of info.
> Search: you ask an engine to help you find what you need.
> Solve: humans and machines collaborate to get things done.

You don’t just search for the local pizzeria. You tell your assistant that you want a large pepperoni and mushroom on your doorstep at 8pm, and it’s there. Tab and tip paid.

Less than six months after launch, Viv was acquired by Samsung, who’s battling Apple for the top spot in the smartphone market. The irony.

And competition will be stiff, with Google Assistant, Microsoft Cortana, and Amazon Alexa all vying to be your go-to assistant of choice.

It’ll be an AI-everywhere world. As Google CEO Sundar Pichai put it:

We’re in the process of embedding ourselves inside a giant, ubiquitous computer. One that will learn from your behavior, and rewrite its own code to accomplish new tasks.

And it’s not just about voice. The next generation of interfaces will be designed for all five senses—allowing people to exchange information in the most natural way for the task at hand.

We’re on the brink of realizing SRI’s original vision: human-like thinking machines that help people realize their potential. It won’t happen overnight, but I wouldn’t fall asleep for too long.

Staying human in an AI world

Our relationship with technology has always been a weird one.

In the early 1800s, the Luddites destroyed cotton looms because they feared the new technology. When computers appeared 150 years later, “computerphobia” spread across the nation. Today, prominent voices like Stephen Hawking and Elon Musk warn of an AI takeover.

Job loss, Terminators, the Matrix, the Singularity. Susan has her concerns too:

Sure, there are concerns to discuss. But I’m optimistic for a positive symbiosis. Machines will continue to touch our emotions and expand our minds. And they’ll help us become more aware of our own humanity. Or as Kevin Kelly puts it:

Will computers ever truly feel human? Will they pass the Turing test? Will they convince the non-believer to fall in love with them? Yes. Yes. And yes.

Siri opened up a new kind of conversation between human and machine—an iconic snapshot along the broader quest to build more intelligent interfaces.

As fate would have it, Susan helped shape Siri. And Siri shaped Susan. But now they’ve parted ways, and the future stands wide open.

Siri is dying. Long live Susan Bennett.


Hey there, wondering where this article came from? It was produced by some people at Typeform, a tool for conversational data collection. Get better data with surveys, quizzes, forms, and polls that look and feel like you.