Developers who worked on Amazon’s new home robot Astro say it's a “privacy nightmare” and will “throw itself down a flight of stairs”. The retail giant claims, however, that the $999 device will offer “peace of mind” to users.
Unveiled on Tuesday, the Alexa-enabled household robot can apparently do everything an Echo device can – including playing TV shows and making video calls – but is also designed to follow people around the house and track behavior using its facial recognition and machine learning systems.
Citing leaked internal development documents and video footage of software development meetings, Vice’s Motherboard reported that the robot forces users to “enroll” the faces and voices of anyone likely to be in the house and conducts “patrols” when “unidentified” people are around, “looking and listening for unusual activity”.
As part of this “Sentry Mode”, the robot will investigate any “unrecognized person” or “audio event” it detects until the feature is disabled. Once it notices a “presence”, it will then apparently approach or follow the stranger and initiate a series of “investigation activities”, including recording video and audio that it then automatically uploads for later viewing.
According to a social robotics document accessed by the outlet, the goal was to make Astro – referred to by its internal project code name Vesta – an “intelligent robot” capable of “some simple but magical interactions with people.”
However, one unidentified developer who worked on the project told Motherboard that the device’s human recognition system was “unreliable at best”, adding that using it as a home security option was “laughable”. The developer noted that the “device feels fragile for something with an absurd cost” and pointed to issues like broken masts – which would not be fixed since there is no way to ship the machine back to Amazon.
They break themselves and will almost certainly fall down stairs in real world users’ homes.
One developer said that pushing Astro as an “accessibility device” was “absurdist nonsense” and “potentially dangerous.”
The outlet quoted another unnamed developer who claimed that it was “not ready for release” and labelled the robot “a privacy nightmare that is an indictment of our society and how we trade privacy for convenience” with such devices.
An additional privacy concern is the robot’s design feature to integrate with Amazon’s controversial Ring products and Alexa Guard, the company’s home security service. Privacy watchdog Electronic Frontier Foundation alleged last year that the Ring app was “packed with third-party trackers” that collected names, private IP addresses and mobile network carriers, among other identifying data.
In response to the criticism, Amazon has said that it had consulted with academics specializing in “computer vision” and “algorithmic bias” when developing the robot’s person-recognition systems. According to one of these scholars, the company was “thoughtful in the design, testing, and augmentation of their approaches as driven by data and feedback in order to minimize bias from their visual ID feature.”
Amazon said negative characterizations of the robot’s performance, mast, and safety systems were “simply inaccurate” and claimed it had gone through “rigorous testing on both quality and safety, including tens of thousands of hours of testing with beta participants.”
Think your friends would be interested? Share this story!
