Home » Business AI » Separating Hype from Reality in Voice Technology
Business AI

Separating Hype from Reality in Voice Technology

Tom Hebner, head of innovation at Nuance Communications, explains why conversational AI needs to first find the problem that needs to be solved.

Tom Hebner

Head of Innovation, Nuance Communications

Voice assistants like Alexa have become ubiquitous. What’s the state of conversational artificial intelligence (AI) today?

Although today we can accomplish simpler tasks that help solve immediate consumer needs — such as requesting a song through a smart speaker or telling your car where to take you via GPS — that’s not a true dialogue. We still have a ways to go before we can really, truly engage intelligently with the systems around us. That’s where the most interesting work is happening when it comes to conversational AI — understanding how we get to a future where we can have those impactful, human-like conversations.

What’s holding us back?

Today, building truly intelligent virtual assistant experiences is often highly manual. While a basic Q&A bot can be stood up easily, and routine rules-based dialog is fairly simple, anything beyond that requires heavy lifting and conversational design experts.

Human dialog is complex, and getting a machine to mimic the way we engage with one another is incredibly challenging. While the recognition part is solved — machines can understand what a human is saying — the next step, processing intent and providing the most appropriate responses, is more complicated.

The application also matters. While consumer-facing voice devices in the home can do things like change the song or set an alarm, a conversational experience for a bank is far more complex. The banking experience needs to be specific and advanced enough that a virtual assistant can help solve a customers’ query. To be effective and valuable, the AI needs to somehow eliminate friction and solve a problem.

How do we break down those barriers?

The market as a whole is being transformed by the potential of deep neural networks and sheer processing power. We are finding ways to automate the build of virtual assistants by harnessing the vast data that exists within certain organizations. This allows us to build high-walled, expert virtual assistants specific to a given industry or job.

The key for brands is understanding the specific opportunity they have and owning it. One of the most used voice-powered devices in the home is the remote control. Why? It is easy and it solves a problem. Instead of scrolling through the guide you can simply ask for your favorite show or network. Organizations will win by defining and perfecting what processes conversational AI will automate, and ensuring it actually makes the experience better for end-users — and doesn’t frustrate them when they can’t have the types of dialogs they want.

Are biometrics being used with conversational AI? 

Biometrics is a really interesting topic in the conversational AI space. Today, biometrics technology is allowing brands to put the days of passwords and PINs behind them, and give customers the chance to validate their identity through only their voice, or the way they tap or text.

In a world where technology is enabling frictionless exchange of money and commerce, security has to play a role. Biometrics will be the future security layer where individuals can engage with AI-powered machines and know they are protected from fraud.

What is different about how you’re working with AI? 

We start with finding the problems we’re trying to solve. Anyone today can build their own machine learning model. Anyone today can build their own chatbot. But will it have value? Will it solve a problem? That’s really where we live.

What’s next in conversational AI?

The hype around AI is significant. The reality is, what’s important is the data and understanding what to do with the data. AI should make lives easier, whether it be interacting with banks and airlines, improving the quality of communication when folks go to the doctor or making the lives of doctors easier by taking a lot of the manual work they have to do out of their hands and doing it automatically.

We have researchers that are taking the latest algorithms and developing new products, like our own Pathfinder, a breakthrough technology that uses machine learning and to increase the conversational intelligence of virtual assistants. That’s why we exist, and why we are making a difference in the world.

Next article