The 2017 State of the Speech Technology Industry: Intelligent Virtual Assistants
But business applications face challenges that Siri, Alexa, and other consumer-oriented systems don’t have, says John Yoon, chief strategy officer at Cypher, a start-up delivering pattern recognition and audio enhancement for mobile communications. Business applications need to have quick response times and extremely low error rates to work well enough to be acceptable in a B2C or B2B environment. IVAs are growing in customer service, but they have yet to reach their potential.
Yoon explains that one of the keys to success in corporate IVAs thus far has been to limit the knowledge behind them. Whereas Siri and the other consumer IVAs need to refer to a database of “everything,” an IVA in a hospitality environment only needs to be able to answer queries related to a specific property or company.
Driving much of the recent growth in IVA deployments has been improved chatbot capabilities, according to Tobias Goebel, director of emerging technology at Aspect Software. “More companies are ready to embrace digital than voice. We are seeing a healthy pipeline in interactive voice response. It’s more efficient to combine non-voice with voice [response] to have the best of both worlds.
“With over 2 billion people using texting and messaging solutions extensively today, offering service over these channels just makes sense,” Goebel says. “Furthermore, it has the promise of saving costs through smart automation while providing a state-of-the-art customer experience.”
Help for People of All Ages
For IVAs to be truly effective, smart automation will be the key. That is where IBM has excelled, blending IVA technology with its Watson cognitive computing platform and bringing in third parties to make specific applications of the technology.
In December, IBM Research and Rice University unveiled the IBM Multi-Purpose Eldercare Robot Assistant (IBM MERA), a Watson-enabled application to help assist the elderly and their caregivers. IBM MERA combines IBM Watson technologies and CameraVitals, a Rice-designed technology that determines vital signs via video recordings of faces, enabling IBM MERA to obtain readings on patients’ heart function and breathing. IBM expects more advances in virtual assistants for the elderly and has opened an “Aging in Place” environment in its Austin-based ThinkLab.
And in January 2016, IBM announced a collaboration with the University of Michigan on an effort dubbed Project Sapphire, an IVA that, much like a human academic adviser, can assist students with scheduling, course and career guidance, and similar advice through conversational speech. When completed, Project Sapphire will rely on 20 years of historical student performance, probabilistic and statistical methods of learning, deep learning, natural language understanding, reinforcement learning, and large volumes of approved, recorded conversations between students and their academic advisers.
Yoon sees virtual assistants advancing further in the medical field as systems get better at eliminating background noise, helping medical professionals, for example, in the routine but critical task of obtaining vital information about patients.
Commercial trucking would also offer a wide variety of applications, Yoon adds. Though some commercial truck drivers use virtual assistants now, such use often consists of personal devices and applications rather than enterprise applications.
Among the challenges with these and other uses are, as noted, eliminating background noise and ensuring that the device knows when it is being addressed, according to Yoon. In noisy commercial applications, the accuracy rate is still far too low to be acceptable. The second issue can be solved by having a naming convention, such as Siri with Apple and Alexa with Amazon.
Collins and Snell expect more use of IVAs in human resources and other administrative operations. Even in hospitality, the technology is finding a home.
At the London Radisson hotel, for example, visitors can use the Edward IVA for many of the same functions as they would a human concierge. Though Edward can’t provide physical tickets for events, it can provide much of the same information and make many of the same arrangements as its human counterparts.
But, experts point out, companies deploying IVAs have to be careful not to take automation too far. After all, there is still occasional value to be derived from real human-to-human contact.
Phillip Britt is a freelance writer in the Chicago area. He can be reached at email@example.com.