Google's primary business is the collection, processing and utilization of user data to help develop and monetize its products, and a post by Google Assistant VP of Engineering Scott Huffman reveals five key things that the company has learned about voice in the past few years. These five actionable insights revolve around what people use voice technologies for and how consumers may want to see the technology evolve in the near future. The insights, in short, are, "Voice is about action", "People expect conversations", "Screens change everything", "Daily Routines Matter", and "Voice is universal".
The first point is one of the most relevant when it comes to AI-based voice programs like Google Assistant and Amazon Alexa. Google's data indicates that Assistant queries are around 40 times more likely to be requests for action than searches. As for the second point, it simply means that people don't naturally do well asking voice assistants to do things by barking predefined commands. Instead, Google has begun teaching its AI systems context as part of its natural language processing program, with training done in large part by the many different ways that people say the same type of command. Setting an alarm, for instance, is something that Google Assistant has heard asked over 5,000 different ways. The third insight is that screens and voice don't necessarily have to be separate; they can complement one another in a number of ways, such as showing a video or pictures of a recipe while spouting off cooking instructions to a user. The fourth point refers to context and routine. People who use Assistant a lot will generally ask for consistent things at similar times on different days, such as the weather in the morning or music when they get home from work. The final point is that voice technology is expanding across the board due to its ease of use. Seniors, children, and even international audiences are increasingly getting on board, making it all the more important to continually optimize the technology for even easier and more intuitive use across a wide variety of scenarios and use cases.
It is worth noting that Huffman frames these five insights as being about voice technology in general, and in many ways, they transcend archetypes like "AI chatbot", "digital assistant" or "voice command system". This is exactly why Google Assistant is built to be a conversational AI that strives to break down what users say or ask and figure out what they want the end result to be.