Speech Technology in the IoT Era
One can't help but have a few visceral reactions when watching Chris Umson, director of Google's Driverless Car program, give his TED talk "What Self-Driving Cars See." Even as you listen to his explanation on how these vehicles work and the problems they solve, mental images of what could go wrong inevitably pop up. After all, can we completely trust our cars to drive us around when we still occasionally argue with Siri?
Therein lies the problem. We are in an era driven by advancements in artificial intelligence, and within our subset of this umbrella, natural language processing, translation, text-to-speech, speech analytics, and other areas are being used in many commercial applications today, from speech-driven IVR and virtual agents to guided resolution for customer support operations. And the results are sometimes spotty.
Since the 1950s, when artificial intelligence (AI) first got serious funding, media and industry hype has fueled the belief that someday there will be an entity such as Hal, the sentient computer in the classic sci-fi movie 2001: A Space Odyssey, or, more recently, the array of artificial intelligence displayed in the computers and "Humanich" androids in the current TV show Extant. Fueling it further are the billions of dollars being pumped into connecting technologies together, not just for the purpose of automation, but to harvest the resulting data to improve how we live. The Internet of Things (IoT), made up of everything from wearables to connected health and the connected home, has led hundreds of companies to work on and deliver products that automate tasks and use and generate data. Indeed, Frost & Sullivan forecasts that the "connected living" concept, comprised of the connected home, connected work, and even connected cities, will be a $731.79 billion market by 2020. Speech and voice technology will undoubtedly boast a major slice of it. Vendors such as Amazon are backing projects to develop voice control of everything from kitchen appliances to home security systems.
But for now, while advancements are being made in creating greater intelligence in the technology and applications we use, most commercial advancements are still point solutions. And we are used to them failing occasionally. So fully entrusting, say, a car to weigh all of the choices that have to be made would seem to have its risks. Will your home automation system let the wrong person into your house, for instance? Will your car swerve in the right direction when faced with an obstacle? A world of fully aware automation, as part of an Internet of Everything (IoE), will take some time to realize. Besides the knotty technological issues within each field, there's the more complex undertaking of tying all that technology together, likely the biggest impediment to Hal meeting the Humanichs.
Furthermore, the intelligence an application brings is typically very specific to it, and without further development, that intelligence can't be generalized to other applications. Thus, IBM's Watson can’t drive Google’s car. Plus, the performance of these applications is tuned to the goals and parameters that have been built into their specific programming. And that's not even getting into the never-ending issues of privacy, security, and ethics to deal with as well.
Still, despite all these obstacles, it is an exciting time to be in speech technology—and in technology in general. Each point solution adds value. As I sit checking stats from my fitness wearable, and I think of seniors being able to live at home longer with monitoring and technology guidance, and I skirt around traffic, getting to my destination using an app, I marvel at all of the positive technological advancements that are and will be making our lives better as hurdles are overcome.
Nancy Jamison is a principal analyst in Customer Contact at Frost & Sullivan. She can be reached at nancy.jamison@frost.com, or follow her on Twitter @NancyJami.
Related Articles
Chatbots and IVAs get all the ink, but RPA applications are vital to a smooth contact center operation
10 Nov 2017