I’ll be the first guy to say that I never expected to write something like this. Yet, technology is now closer than ever to the fabled singularity, as the new Duplex feature in the Google Assistant app shows.
During Google’s I/O keynote 2018, the CEO of the company, Sundar Pichai, introduced the world to the feature. Duplex, as Mr. Pichai would call it, is the new feature of Google assistant that allows the app to call and set appointments for you. To the average viewer, that doesn’t seem anything special. Anyone with any technological expertise however, would be stunned. Google has produced an AI which can speak with the most natural human voice anyone has heard of.
The new feature lets the user set an appointment with a restaurant or any other kind. The app first will look for a way to make the appointment online, but if that’s not possible, it will call. Sundar showed off the AI’s human-like intelligence in a call to a salon. In it, the AI acts very human, making the sounds one might make when impatient. The AI also can read the voice of the person it talks to. Sundar Pichai explained this to be a effort of three pronged research. He said, “The google assistant seamlessly makes a call for you in the background.” He would add, “The amazing thing is the assistant can actually understand the nuances of conversation.” He would also reveal that this is the end result of their research in natural language understanding, deep learning and text-to-speech. The assistant, after finishing, would notify the user of the appointment being set.
The first question that comes to mind is that on the heels of the Cambridge Analytica scandal, is this the right move? Google has a fairly squeaky clean reputation but in 2018 many people have already begun to trust tech company’s less. The Google Assistant already needs many permissions upon installation. Some of these include permission to use the microphone, permission to use the internet and permission to knowledge of current location. After this feature is implemented, the app will surely require more permissions from the user. More permissions, means more data to google. That may not be a good thing for the user, as Facebook has shown.
What’s more stunning is how it’s virtually impossible for anyone to know that the Google Assistant was not human. In fact, the voice seems to be so impeccably human, that they even handled much stranger voice calls by reading the context. I’ve seen enough movies to know that this is at least a little shady.
The technological singularity might be sooner:
My worries do not stop at deception. With such massive strides in AI technology, it’s clear that we are closing in on the singularity of technology sooner than predicted. For the uninitiated, the technological singularity is the hypothesis that the invention of artificial superintelligence will abruptly trigger runaway changes to human civilization. We think of movies like Matrix when we think about the negative side of that. While AI has a long way to go before reaching the singularity, it’s still made huge leaps in the last few years, epitomized by the new feature Google is giving us. Many futurists and scientists alike agree that runaway AI development may end civilization itself if left unchecked.
There are also other kinds of ethical worries. For example, this AI may replace service workers in the near future. In creating the most human-like AI in history, Google might have killed the service industry, or at least severely damaged it in the long run. Automation has been a worry for a while and now it seems no company will need a receptionist ever again.
What this could mean for us:
For the regular Bangladeshi, this won’t change much- for now. Google will surely try to have this technology in as many voices as possible, and Bangla would be a language that will have one eventually. Even before that, it would severely damage the call centers and other service based industries once they start to use this AI for commercial uses. This would also provide Google with even more data from us, as most of us already use it for almost everything. Bangladesh in general is behind the curve on internet safety, and this could be heavily exploited by Google as they could sell that data to third parties. That is assuming they are an equivalent of Facebook. They may not end up selling the data, but the chances remain.
There are many concerns against Google’s new announcement. We mustn’t forget that no technology is inherently evil. Google has made tremendous leaps in the development of AI, which I would like to applaud. Still, it would be dangerous to think of this new development to be harmless. In the end, the best thing we can do is try to monitor how much of our data is being tracked. Different industries also need to make sure this new technology doesn’t cost thousands of jobs.