Founder and Facebook CEO Mark Zuckerberg has a habit of setting personal-growth challenges and goals every year. In 2010, for example, he studied the northern Chinese language, and in 2015 - read two books a month. In 2016 Mark decided to use #Artificial intelligence staying at home. Thus, Blasting News presents impressions from the creator who discussed Jarvis in a video.
Zuckerberg announced that he planned to build an AI system using Facebook tools
"In 2016 I decided to create a simple artificial intelligence like Jarvis from "Iron Man. I would like my personal assistant to control temperature and safety in the home, be responsible for managing a variety of devices, including my preferences in music, my tastes and desires. However, to invent such a "toy" was harder than I thought," he said
As a result, the creation of such a smart electronic assistant took about 100 hours. Jarvis is written in Python, PHP and Objective C. But now this newly-made butler controls light, cameras and other devices in the home, and also opens the door and prepares toasts. Moreover, another great feature - on the first request Jarvis threw Zukcerberg a clean T-shirt.
"Of course, I needed a butler's voice. Robert Downey Jr, who played a major role in the "Iron Man" offered to voice him. However, in the end, I chose the voice of Morgan Freeman."
In the video he wnet on to say that to address the system commands, you can use voice applications for iPhone, surveillance camera or a bot for Facebook Messenger. That is Jarvis takes commands both in oral and written form. Thanks to that option the owner has access to the system, regardless of where it is located. By the way, to control the assistant by voice, we had to create a separate application. It continuously listens and analyses everything that happens around, and determines when it concerns to the butler.
Recordings are processed directly on the server and converted to text. With it, the system can work in the future. In addition, Jarvis sends me messages, images from the cameras and reminds about affairs.
During the implementation of the project were also tough cases
The devices in my house use different interfaces. There are no universal API and performance standards for all "smart" devices. So I had to use reverse engineering techniques. Thus it became possible to manipulate all devices from a private centre.
There are plenty of issues faced by home helpers - for example, recognition of movements on the video. The solution was to create a server that constantly identifies images on the camera.
It was also difficult with the music. There are countless music albums, artists and songs. In all this diversity there are similar words. The more accurate a system context is, the better it understands the commands.
Understanding the context - is a valuable feature for artificial intelligence. When two people from the two separate rooms asked to turn on the light, then the system perceives it differently. If you ask the system to include music, without naming the room, it is necessary to determine the location of a person.
I will present Jarvis to the public next year in September. Today, the application is developed for iOS, in the nearest future - for Android. It is also planned to work on studying algorithms. In the long term - to understand how the system could be used in other homes.
Creating a home assistant helped me to understand in what status is artificial intelligence nowadays."