Nuance CTO: Conversational AI is the 'next big step'

Conversational AI has steadily grown more advanced over the past several years. Nuance CTO Joe Petro explains why the vendor is refocusing on the technology.

Advances in AI, natural language processing and natural language understanding, sparked by cloud computing and more powerful hardware, have powered a new wave of intelligent virtual assistants and chatbots.

The same advancements over the last several years have also enabled business users to query their data in natural language using voice and text, and consumers to use plain language to ask smart devices to run various tasks.

With that in mind, longtime conversational AI vendor Nuance refocused its core business on enhancing and creating new products that use NLP, NLU and voice to better power healthcare, financial services and customer care applications.

Last month, the company, based in Burlington, Mass., announced plans to sell its Health Information Management (HIM) Transcription business and Electronic Health Record (EHR) Go-Live Services business to a new company, DeliverHealth Solutions LLC. Nuance said it will be a minority shareholder of DeliverHealth, with the transaction expected to be completed in early 2021.

In a Q&A, Nuance CTO Joe Petro discusses the planned sale, as well as the company's refocused efforts on building out its core conversational AI business.

How will the planned sale of Nuance's HIM and EHR businesses further Nuance's goals?

Joe Petro: If you look at our behavior over the course of the last two years or so, we've been really focusing the portfolio to support all of our plans, aspirations and current business commitments in the area of conversational AI.

Joe Petro, CTO, NuanceJoe Petro

If you look at HIM in general, it's a very services- and labor-intensive business. It is key to hospitals and institutions, but it's not part of the modern era in terms of all the various cloud things bringing to bear on all the problem that we're trying to solve. So, it was one of these puzzle pieces that kind of fit but was not in the center of gravity in terms of how we're moving the company forward.

This allows us to continue to commit to the software-based models that we've got on the market, and all the stuff we're bringing through the cloud on conversational AI.

What are some of the Nuance conversational AI products?

Petro: One of the big ones in this mix, of course, is our DIY conversational AI tool. It's a big part of our core technology platform. When we think about conversational AI, we're really talking about speech, natural language processing, as well as dialogue. These are all of the various technologies that allow you to model a very natural conversation between a human being and a computer system.

On the enterprise side, you can think of it as we weaponize all the various channels coming into a large company. It's very much kind of a DIY development environment that allows us to deploy our services, and for either our clients or ourselves to deliver the outcomes that they're after. All of that conversational AI stuff, that's really all about our client's brand, and how they actually manage the brand and manage the touchpoint with their customers.

We've been really focusing the portfolio to support all of our plans, aspirations and current business commitments in the area of conversational AI.
Joe PetroCTO, Nuance

In addition to that, we've also introduced a variety of products around voice biometrics and security, powered by AI-based, neural-based voice identification technology. We're also bringing a lot of fraud detection capabilities to the market as well, which combines well with the voice biometrics software.

In terms of the fraud detection capability, we've got something called behavioral biometrics, which also comes into play. If you're calling into a company, we can, with behavioral biometrics, identify if you're calling in with your usual cellphone, using your normal phraseology and speed of speech. We can use that information to detect fraud. We actually just released one of the first instantiations of biometrics protocol, Gatekeeper.

We also have Dragon Ambient eXperience, a product that listens to the conversation between a patient and a physician through either a smart device in the room like a smartphone, or an actual permanent physical device in the room. The product runs it through AI, a gigantic neural network with something like 75,000 parameters in this particular case.

The product diarizes the conversation, creating a fully formatted, summarized, clinically accurate document. The document is then sent back to the physician.

Nuance sees conversational AI as a big part of the future. Why is that?

Petro: It's the next step and kind of our journey, right. Ten years ago, it wasn't very common to talk to a phone or a computer, but since then, we've made a lot of strides. That's all based on speech technology, which started off with mathematical and statistical models, and has since moved on to AI and neural nets. In parallel with that, natural language processing is coming to its highest capability, where you can start to understand what somebody says by extracting facts and evidence.

So, if a physician, for example, says that they're going to place an order for amoxicillin, 250 milligrams by bottle, natural language processing can understand that amoxicillin is the drug, while the dosage is 250 milligrams. It might also now realize that the physician is not only talking about the drug, but they're actually trying to place an order by analyzing intent.

Conversational AI really enables the next step in, in kind of our natural interaction with computers by voice, and it's far better, of course, than doing the point-and-click type of thing that we've all done over the last 25 years since we've been heavily using computers. For us, it intersects directly with the trajectory that we're on in terms of the intelligent engagement with systems, so it's an exciting time.

Editor's note: This interview has been edited for brevity and clarity.

Dig Deeper on AI business strategies

Business Analytics
Data Management