About a month ago at the I/O 2018 Conference, Sundar Pichai unveiled Google’s latest attempt at revolutionising the world. And this time, more than ever, the reveal generated discussion about the ethics of modern technology. The product? Google Duplex. The latest advance in AI that aims to mimic human conversation to almost pinpoint accuracy. The goal is to make it easier for us to make bookings and reservations at any outlet by skipping the human interaction required, instead relying on AI to carry it out for us. Watch the video below from 35 minutes in.
If that hasn’t blown your mind, I am not quite sure what will. In order to reach the level of sophistication showcased, Google obviously had to train their model repeatedly and get it to adapt to various situations. What makes the software so lifelike and accurate is in how it is able to capture the intricacies of human conversation: the pauses, the ability to keep track of points brought up a few moments ago, enunciating with different tones and quite possibly injecting. The voice samples, unlike the one’s used in Google Now/Assistant’s voice search, are a lot more cohesive and do not possess slight stutters or gaps between phrases. This has been made possible with the help of Wavenet, which uses a different approach to sampling.
A lot has been made about the ethics behind this latest innovation. Shouldn’t people have the right to know whether they are conversing with an AI or a human? This comes in the wake of the Moucheng Controversy, where Chinese dating apps were found to be defrauding men by using female chat bots. How long will it be before we use AI to substitute talking with our friends?
We need to keep in mind that even though Google intends on pushing the software ‘’within a few months’’, the AI behind it is far from perfect. The example we were shown was a highly trained and curated model for specific scenarios. These models have to be tested thousands of times and need to be retrained in order to execute them perfectly. Besides, this isn’t the first time Google has unveiled something ground breaking which failed to live up to expectations.
Jump back to a year ago – what was the latest release by Google making news? The Pixel Buds. They had the ability translate languages in real-time and would finally bring to reality what Douglas Adam’s envisioned with the Babel Fish. Elimination of Language Barriers. The reveal followed a pattern similar to the one this year, with an example which defied any and all expectations. But as we now know, the buds were released a few months later and failed to live up to the hype. A product possessing functionalities which were exaggerated beyond measure.
Two other projects which fell by the wayside were Project Ara and Google Glass.
Ara was supposed to be the next step in mobile phone technology. A truly modular smartphone that could replace parts as and when required. ‘’Oh, is your camera sensor all battered up? That’s alright. Switch for a new one without any hassle. Do you need a boost in battery for the long trip abroad? Here’s a 4000maH battery to switch in for your present one.’’ But various issues constantly plagued the project which made it impractical and didn’t pass the testing stage. Glass was supposed to be the new accessory that went along with your smartphone, being able to take photos, record videos, make calls and what not. Once again, the project proved to be unfeasible and the project was scrapped.
You might be wondering what I am driving at? Discussing about the negative effects of technology should be carried out whenever something revolutionary is revealed. But we should not blow it out of proportion. Media in today’s world have this tendency to go all out and create unnecessary panic amongst the masses, especially when the product hasn’t even been launched yet.
Google experiments on various moonshot ideas and they hope to make use of their better aspects. The image processing tool from the Google Glass is now used to help identify and label objects. Down the line we might just be able to break that language barrier with the help of a third party developer working on what Google has done so far. And by the time Duplex reaches the complexity required to converse naturally with humans, we would have hopefully come up with some rules and restrictions on where to use them.
Duplex might end up being one of Google’s many failed projects. But only with brilliant experiments like these can we come up with ideas that grow into something new.
Maybe it’s time for Google to take a step back. They have been so focussed on an all AI future that they seem to have lost focus on their Do No Evil policy. Sort out the mess with Pixel Buds, decide on a plan of action for the drone situation (edit:Project Maven has been cancelled- https://www.theverge.com/2018/6/1/17418406/google-maven-drone-imagery-ai-contract-expire). Google Fi is a brilliant initiative, and we haven’t heard too many updates on that front either. Juggling so many balls is one thing, but there comes a point where you say no more.
At the end of the day, we need to keep calm about new developments in the field and remember: Don’t Panic
alotofrandomcrap 