17 Jul

You Shouldn’t Fear Amazon’s Alexa

Why paranoia about digital voice assistants is overblown

Star Trek vs. George Orwell’s 1984. No two cultural touchstones better illustrate our diametrically opposed feelings about voice systems.

Star Trek’s always listening, ever-helpful Computer represents the highest ideal of a digital assistant, while Orwell’s Telescreen, with its “Big Brother is watching” messages, was emblematic of our darkest fears. With each passing year and digital assistant breakthrough, we vacillate wildly between these two perspectives.

Bloomberg recently reported that Amazon was employing thousands of humans to comb through utterances and transcribe what we say to Amazon’s Echo-based voice assistant. The revelation broke an unspoken agreement between Amazon and the millions of Echo and Alexa-enabled device owners who assumed that only algorithms, not humans in a back office somewhere, would be analyzing our words for meaning.

Amazon isn’t Big Brother, but its 613,000 employees and $232.89 billion in annual sales revenue puts it at nation-state scale, raising concerns about a company with that much power listening in on our private conversations

Upon learning of this, many of us, myself included, paused to remember those moments when Alexa responded even when we didn’t utter the “Alexa” watchword. What did we say to prompt it, and what was Amazon doing with the information?

In reality, not much. Amazon didn’t deny its use of flesh and blood people to dip into the Alexa conversation stream, plucking words and sentences for its own purpose. However, those purposes were, as Amazon tried to make clear in its official response, more benign than anything Orwell might have imagined: “We only annotate an extremely small sample of Alexa voice recordings in order [to] improve the customer experience. For example, this information helps us train our speech recognition and natural language understanding systems, so Alexa can better understand your requests, and ensure the service works well for everyone.”

It went on to say that those transcribed conversations are not connected to any personally identifiable information.

Amazon has probably heard and transcribed some pretty bizarre and very personal conversations. So what? Amazon isn’t going to do anything about what it hears.

Amazon’s interest in what you’re saying is in direct proportion to how what you say relates to activities in the company’s ecosystem. Alexa connects to some 80,000 skills, but the core set — the ones you use most often — are connected to Amazon’s shopping, music, and the company’s growing collection of smart devices. Does it have to “listen” all the time to connect those services to your needs? Yes. But so do Siri, Google Assistant, and Samsung Bixby. They all listen for their watchwords (just as Star Trek’s voice system always listened for “Computer”). Nothing that the system “hears” before that is collected, but what’s said after you utter “Alexa” is.

Obviously, Amazon could have done a better job of communicating the possibility that some of your conversations could end up in human hands. On the other hand, at this stage, no voice system can improve without a combination of algorithmic machine learning and human analysis.

Amazon is far from alone in its desire to perfect these systems. Apple, Amazon, and Google have spent years improving both the voice quality and conversational- and context-aware capabilities of their voice systems. Last year, Google’s Duplex carried on a conversation with a hair salon, even booking an appointment, all without the person on the other side being aware they were talking to a computer — which predictably enough freaked some people out.

Meanwhile, more typical digital assistants like Alexa, Siri, and Google Assistant still fall far short of the voice assistant ideal envisioned by Star Trek in 1966.

No other sci-fi show has had such a lasting impact on today’s tech leaders, and if you want to blame something for the proliferation of voice assistants and the way they work, there is no better target than Star Trek. All these middle-aged programmers grew up watching one iteration or another of the sci-fi franchise. The concept of an always-listening computer spread like a contagion. One of the chief architects of Google’s Knowledge Graph, former Google Senior Vice President Amat Singhal told me in 2012 that it was his dream to build the Star Trek Computer. And in a recent letter to shareholders, Amazon founder and CEO Jeff Bezos wrote, “The vision for Echo and Alexa was inspired by the Star Trek computer.”

Was Star Trek’s Computer so special?

While it’s not clear exactly what inspired Gene Roddenberry to introduce Star Trek’s Computer, we do know that he deemed it important enough to give the voice assistant his wife Majel Barrett’s voice. Her voice continued to be used for the Computer even after her death.

Barret didn’t voice all the Computer versions, though. In one original series episode, Captain Kirk, played by William Shatner, engages in a lengthy back-and-forth with the Library computer. It’s just the kind of conversation, one that never pauses to reset context, that we all dream of with our current voice assistants.

All those times Shatner and his castmates spoke to Computer, did they think about the implications for the future? Were they, perhaps, wishing they had such an assistant of their own?

“Isn’t that what sci-fi is, taking something that exists today and projecting a possible future? Did I know? No. It’s the writers that projected the future. I just read the lines given by them,” Shatner told me via Twitter DM.

Roddenberry’s future was an optimistic one, one tech companies wholeheartedly embraced. But consumers today are torn. As Bezos pointed out in that same letter to shareholders, no one at the time was asking for Echo: “If you had gone to a customer in 2013 and said, ‘Would you like a black, always-on cylinder in your kitchen about the size of a Pringles can that you can talk to and ask questions, that also turns on your lights and plays music?’ I guarantee you they’d have looked at you strangely and said, ‘No, thank you.’”

For consumers, an always-watching and listening system will always smack too much of Big Brother, even if it’s driven by a dispassionate algorithm. It’s the job of Amazon, Apple, and Google to reconcile the promise of the Star Trek Computer with the threat of Big Brother. In this area, Amazon clearly still has some work to do.

Stop complaining and enjoy the fact that you can talk to a computer.

But let’s say Amazon and other voice assistant developers get it right. Where would such an advanced voice assistant lead us?

Remember the 2013 movie Her starring Joaquin Phoenix? Its depiction of a lonely man falling in love with his digital assistant (voiced by Scarlett Johansson) was unsettling, not only because of the cross-silicon-human relationship but because of its inevitability. (Spoiler alert: The voice assistant eventually “leaves” her human to live with other A.I. in the cloud.)

If humans eventually develop meaningful relationships with digital beings, it will happen because of all those years of transcribing and training worked. So should we be concerned that Alexa and Amazon are listening to everything we say? No. Sure, Alexa pipes up even when we don’t explicitly say “Alexa” because it thought it heard its name, and yes, that means Amazon has probably heard and transcribed some pretty bizarre and very personal conversations. So what? Amazon isn’t going to do anything about what it hears.

This is the price of innovation. If you don’t want Amazon hearing what you say at the dinner table, turn off the mic or throw away your Echo. Otherwise, stop complaining and enjoy the fact that you can talk to a computer and that it can answer almost any question you throw at it.


Author: Lance Ulanoff

Source: One Zero

Original Article

Leave a Reply