What will the world look like when our primary means of interaction with technology is voice, and no longer touch?

by Steve Tam, VP Marketing @Snips

For the past ten years we’ve been tapping palm sized screens with oversized fingers to control the technology around us. From playing that last single on repeat, to changing the thermostat we’ve clicked the phone on > opened an app > and navigated a unique interface to take control.

The fact that you can almost blindly add a photo to your Instagram feed, or book a restaurant on OpenTable is impressive, but it took some muscle and mind memory to get there. What’s even more impressive is that every app’s unique interface requires some mental capacity to learn and remember.

So what happens when we have to learn not just how to use a multitude of apps — each with their own unique UX — but a multitude of devices? The number of connected devices per user is already beginning to hockey-stick — everything from our coffee machines to toothbrushes will be connected before we know it. And they’ll all run on apps. How many more app interfaces will we be able to handle before things become overwhelming and prohibitively chaotic (if they aren’t already?!) Enter our new interface savior — voice. Voice is our most natural interface, it will democratize access to the connected world while relieving the mental load of operating touch screens. Even my grandmother can yell at the air conditioner!

We believe that voice will become our primary means of interaction for much of the world around us.

And while today it might be a bit annoying to have… to… speak… like… this… to Siri, and you might have forgotten that Alexa has been sitting in your kitchen for months, the promised land is near!

Voice interactions rely on two main technical components: the ability to translate sound into text, and the ability to translate text expressed in natural language into a standardized object a classic algorithm can act upon. Those two problems are perfect playgrounds for machine learning, that simply makes solutions more scalable and more perfectible than traditional methods.

What we’re observing is that the key players of the industry are progressively unlocking blockers one after the other: scaling to new languages, to new intents, to more challenging conditions: distance, noise, acoustics, and beyond.

Machine learning is being used by all the key players of the industry — but it is lacking one fundamental element. Privacy. Outside of the work we’re pioneering at Snips, all other connected devices are lacking Privacy. Voice interactions will not improve at the rate they should if doing so requires having always-on microphones feeding all of the sound they record to centralized servers. This is just too much of a hazard and privacy risk. Privacy is what singles out Snips in this moment — makes us different, and better. However, it’s so important for others to join the fore.

Connected devices that lack privacy are in danger of lacking potential.

We expect that consumers will demand a solution that offers the same level of quality as the state of the art, but that runs on-device. That means that voice interfaces will strictly rely on local processing, which is a fundamentally safe approach to privacy. We call this privacy by design.

We’re truly excited about the possibility of keeping our phones in our pockets a little longer every day, and starting to speak to the world around us.

Join 30,000+ people who read the weekly 🤖Machine Learnings🤖newsletter to understand how AI will impact the way they work.