Facebooktwittergoogle_pluslinkedinmail

AI vs ASR Main Image

A Place for AI

When most people hear Artificial Intelligence, they tend to think in terms of Hollywood and famous AI characters like HAL9000 and Skynet. Business applications of AI are nowhere near that sexy, but, on the bright side, that also means that our robotic overlords aren’t looming on the horizon just yet.

The reality of AI at this point revolves less around machine self-awareness and more around predicting outcomes by harnessing incredible computing power. Business processes and communications all stand to benefit from the continued evolution of AI. Interactive voice response (IVR) definitely belongs on that list of beneficiaries.

Before diving into specific processes that can make the most out of AI, it’s worth having some additional context in order to understand why AI is going to make a splash in the world of voice automation.

A New Understanding

IBM’s Watson is synonymous with artificial intelligence these days–and it’s not just because Watson cleaned up on Jeopardy! In the voice communications space, Watson’s ability to handle natural language is a big reason why a lot of people, including the engineers at Plum, are excited for the possibilities Watson integrations hold.

We’ve discussed the time and money that goes into natural language speech recognition software, like Siri, Alexa, and Cortana, in these pages before. The reason behind this is that automatic speech recognition (ASR) and natural language understanding are built on two different assumptions.

ASR is all about expected answers. This means asking specific questions and anticipating, not only what people will say in response, but also all the different ways in which they can say that. Think of how many ways there are to say “yes”– yup, yeah, sure, uh huh, ya, you bet – these are just a few of the dozens of ways to indicate agreement or assent. When you consider an entire phrase, the workload increases exponentially.

The collection of expected responses constitute a grammar. It’s necessary to constantly tune grammars to ensure they catch every possible utterance variation. The more variances the software encounters, the more it must process. It’s not hard to see how this can quickly become time and cost-intensive work.

With AI and natural language understanding, the software focuses less on expected answers and attempts to determine the intent of what is said. Therefore, it’s possible to do a whole lot more with live voice communications without the insane costs associated with compiling and tuning grammars.

IBM’s Watson provides API access to its natural language understanding technology and companies can quickly and easily tap into a machine learning capabilities.

The Road Ahead

The ability to discern intent without involving a human opens up all kinds of possibilities for business communications (whether that’s internally with employees or externally with customers) and customer service. Incorporating Watson technology into IVR applications combines the best aspects of self-service in the digital realm with the immediacy of real-time voice communications.

Now that we’ve established a bit of context for what AI like Watson can do for voice communications we’ll turn to some examples of how companies might actually use that technology. Stay tuned for more information and ideas in these pages.

(Visited 205 time, 85 visit today)

About   

Jason Myers stitches together letters and words into cogent thoughts as the Copywriter at Plum Voice.

    Also find me at: 
  • linkedin
  • twitter