I was visiting a friend a few weeks ago when he started bragging about how he set up an Amazon Echo in his home office. “Alexa, what is the weather outside,” he volunteered unfettered—even as I could see the sun shining brightly out his window. In a few seconds, a rather pleasant computerized woman’s voice filled the room confirming my observation.
“Listen to this,” he continued. “Alexa, play Elton John’s ‘Candle in the Wind’.” A few moments later, the song came on.
It was getting irritating, so I decided to have a little fun. Before my friend could stop me, I commanded Alexa to place an order for a brown, four-shelf bookshelf. “Your order has been placed,” Alexa responded.
The next five minutes were frantic. My friend desperately fluttered on his keyboard trying to find customer support, but the answer was obvious. “Alexa,” I said sternly, “cancel the bookshelf order.” She confirmed.
Google’s co-founder, Larry Page, once described the perfect search engine as a machine that “understands exactly what you mean and gives you back exactly what you want.”
If he’s right, then the intersection of artificial intelligence and voice recognition is the pivot point. Google, the largest purveyor of search results on the Internet, has invested heavily—both dollars and engineering prowess—in data mining and artificial intelligence. The result is a technology likened to the talking computer on Star Trek, or a souped-up Siri, Apple’s voice-controlled virtual assistant. But Google claims its Google Assistant will be ever-more powerful than Apple’s Siri, Microsoft’s Cortana, or Amazon’s Alexa.
Sundar Pichai, Google’s chief executive, says that machine learning is at a point where a virtual assistant is all we need to solve all our information-related needs. Google Assistant will learn our habits, our likes and dislikes, and have access to just about all our confidential information. It will have the processing strength to understand and contextualize what we want and how we want it. It will book a trip, buy a coat, order a pizza, and make an appointment with a favorite hairdresser.
Building something better than Alexa, Siri, and Cortana is ambitious, but as Henry Lieberman, a pioneer of human-computer interaction at MIT’s Media Lab, told Associate Editor Alan Brown in this month’s cover story, “Language will become a means—not to help users understand a product more easily, but to have the product understand its users.”
The impact of harnessing the power of voice—and cognitive—recognition on product and systems design is still unclear. But we’ve seen significant strides in deep neural networks, referred to as deep learning. These are software constructs that enable machines to teach themselves how to recognize complex patterns. They have also greatly improved speech recognition.
Responding to public concern over the impact of machine learning on robots and intelligent systems, including factory automation and self-driving cars, a consortium of technology companies, including Amazon, Facebook, Google, IBM, and Microsoft, recently formed the Partnership on Artificial Intelligence to Benefit People and Society. Its focus is on ways to protect humans in the face of rapid advances in AI, and the potential for government regulation of the technology.
Sure, Alexa understood my command to cancel my joke order for the bookcase—that was trivial. But it’s critical that the engineering community recognizes the importance of building AI into the design of technologies in a way that doesn’t violate ethical mores. That’s no laughing matter.