This May, Google has given its digital assistant a brand new ability: to make phone calls on behalf of the user. The artificial intelligence software present on pretty much every modern-day smartphone running the Android operating system has been demonstrated to book a haircut for the owner and to make a restaurant reservation.
At no point during the reservation did it identify itself as an AI, and it has even added sounds typically produced by a real human while speaking, like “um” and such, to make the call as natural as possible. The technology called Google Duplex will “save time for people and generate a lot of value for companies”, Google chief executive Sundar Pichai said upon demonstrating the technology at Google’s annual IO conference. But do these benefits offset the risks of building AIs indistinguishable from human beings?
The future of cold calling
“OK Google, book a table at Larry’s for 7 PM tomorrow” might be a phrase we might be able to use pretty soon. When Google Assistant hears us, it will be able to find “Larry’s” either in our phone book or in its own vast database, make a call, book a table in our name, and put the appointment in our calendar. This will be very useful for those with a truly busy schedule who can’t spare the time – or constantly forget because of their busy schedule, and will help Google Assistant to behave a lot like everyone’s personal secretary.
On the other hand, end-users will likely not be the only ones to make use of this new technology. Most likely, there will be a version of the technology destined to be used by companies that will be able to abandon the current practice of maintaining call centers to promote their products and services through cold calling and for those providing phone-based technical support to their customers. Just like bots are responding to various requests on Facebook Messenger, the customer support representative of the future might also be a bot – but this time, one that speaks like a human.
Will people trust an AI?
At first, they most likely won’t. An AI personal assistant posing as a human being – going as far as using “fillers” to make its speech more natural – will probably make many people uneasy when the technology will be first rolled out to the masses (if ever, that is). A legal obligation for an automated caller will probably be introduced, making it mandatory for Google Assistant (and all of its epigones) to identify themselves as AIs before engaging in a conversation with a human (Hello, I’m an AI calling on behalf of /User/). Slowly but steadily, AI callers will blend in with the world, finding their place in the already overloaded tech ecosystem.
Considering that at their current state, AI assistants are merely a more advanced interface for a massive database, people will likely trust them as much as they trust a search engine. There are many out there who hate to make phone calls – they will probably embrace such a technology much easier than others. After all, these assistant programs are not truly intelligent – they can’t make decisions on their own and they are not capable of feelings and creativity.
Voice-controlled digital assistants are already pretty widespread – their ability to make appointments by phone will probably make them more attractive for users, sparing them another inconvenient chore. People will likely embrace such a new feature pretty easily.