Conversational AI Is Becoming More Practical
After a two-year hiatus, in-person conversational speech and AI shows are back! The Conversational Interaction (CI) Conference, held April 12 and 13 in San Jose, Calif., was organized jointly by Bill Meisel’s TMA Associates and LUI (Language User Interface) News and AVIOS, a nonprofit speech technology professional society. While speech technologies and artificial intelligence have a growing presence at other industry shows, such as Enterprise Connect, which was also held in April, the CI2022 conference is one of the few events highly targeted at practical use of speech and AI, and it acts as a place to track trends and industry sentiment.
The event did not disappoint and showed that as an industry we have come a long way from the early days, when the initial CI conference was called Mobile Voice. Meisel nailed it when he kicked off the conference with the keynote panel, “What’s the Status of Conversational Technology Today and Where Are We Heading,” which he moderated and which put the technology’s progress on full display. The panel’s consensus: We have moved beyond simple chatbots to making self-service applications more intelligent, achieved more with transcription and contextualization of interactions, and developed more complex levels of analytics to glean sentiment and intent. These trends certainly mirror what we are seeing in customer surveys at Frost & Sullivan. For example, its 2022 IT Decision-Maker (ITDM) Survey results show that self-service solutions infused with AI will be the top channel for investment (see chart). While this trend was ramping up before COVID-19, the pandemic accelerated investments.
The survey also found that AI will be deployed in some form or another by almost 80 percent of organizations by the end of 2022, with intelligent virtual agents (IVAs) and speech technologies as top investment priorities.
The challenges faced by implementing conversational AI were a central theme throughout the conference. Chief among them were not repeating the mistakes of the past, such as we did with poorly designed IVRs, and the need for best practices. Integrating knowledge management to get customers to the information they need, along with shortening the path to get there, is a top design priority. Yet when self-service attempts get the desired results, the bot or IVA should “fail gracefully,” as one presenter pointed out, so that customers don’t get frustrated, and the system can get them back on course while always providing an escalation path if necessary. Others top challenges noted include making sure bots are accessible, inclusive, and empathetic, as well as better able to handle multiple languages, accents, and cultural nuance.
What Is Next?
In play now is an increase in the number of vendors supplying digital humans, often called avatars, to increase application effectiveness, customer engagement, and enhanced branding; they are getting better all the time. A number of IVA providers are deploying this emerging digital workforce in brick-and-mortar businesses, such as in life-size digital displays (kiosks) and vending machines; in car navigation systems; and, most commonly, on websites. Two areas emphasized at the conference were the advances in making the characters talk and gesture as humans do, and the presentation of customer use cases, more so than in years past, across a wide spectrum, from chatbots and contact center examples to large scale deployments of speech apps to improve employee experience.
While much work needs to be done to bring mainstream conversational AI to the masses, it was nice to see the conference highlight the positive momentum and work being done to increase speech and AI adoption and bring more intelligence and human traits, such as empathy, to applications.
Nancy Jamison is a senior industry director in information and communications technologies at Frost & Sullivan. She can be reached at firstname.lastname@example.org, or follow her on Twitter @NancyJami.