Google just gave a stunning demo of Assistant making an actual phone call
Onstage at I/O 2018, Google showed off a jaw-dropping new capability of Google Assistant: in the not too distant future, it’s going to make phone calls on your behalf. CEO Sundar Pichai played back a phone call recording that he said was placed by the Assistant to a hair salon. The voice sounded incredibly natural; the person on the other end had no idea they were talking to a digital AI helper. Google Assistant even dropped in a super casual “mmhmmm” early in the conversation.
Pichai reiterated that this was a real call using Assistant and not some staged demo. “The amazing thing is that Assistant can actually understand the nuances of conversation,” he said. “We’ve been working on this technology for many years. It’s called Google Duplex.”
Duplex really feels like next-level AI stuff, but Google’s chief executive said it’s still very much under development. Google plans to conduct early testing of Duplex inside Assistant this summer “to help users make restaurant reservations, schedule hair salon appointments, and get holiday hours over the phone.”
Pichai says the Assistant can react intelligently even when a conversation “doesn’t go as expected” and veers off course a bit from the given objective. “We’re still developing this technology, and we want to work hard to get this right,” he said. “We really want it to work in cases, say, if you’re a busy parent in the morningand your kid is sick and you want to call for a doctor’s appointment.” Google has published a blog postwith more details and soundbites of Duplex in action.
“The technology is directed towards completing specific tasks, such as scheduling certain types of appointments. For such tasks, the system makes the conversational experience as natural as possible, allowing people to speak normally, like they would to another person, without having to adapt to a machine.” Google envisions other use cases like having Assistant call businesses and inquire about their hours to help keep Maps listings up to date. The company says it wants to be transparent about where and when Duplex is being used, as a voice that sounds this realistic and convincing is certain to raise some questions.
In current testing, Google notes that Duplex successfully completes most conversations and tasks on its own without any intervention from a person on Google’s end. But there are cases where it gets overwhelmed and hands off to a human operator. This section on the ins and outs of Duplex is very interesting:
The Google Duplex system is capable of carrying out sophisticated conversations and it completes the majority of its tasks fully autonomously, without human involvement. The system has a self-monitoring capability, which allows it to recognize the tasks it cannot complete autonomously (e.g., scheduling an unusually complex appointment). In these cases, it signals to a human operator, who can complete the task.
To train the system in a new domain, we use real-time supervised training. This is comparable to the training practices of many disciplines, where an instructor supervises a student as they are doing their job, providing guidance as needed, and making sure that the task is performed at the instructor’s level of quality. In the Duplex system, experienced operators act as the instructors. By monitoring the system as it makes phone calls in a new domain, they can affect the behavior of the system in real time as needed. This continues until the system performs at the desired quality level, at which point the supervision stops and the system can make calls autonomously.