Dear subscribers,
It is one thing to hear about algorithm biases.
It is quite another to have the demonstration in front of you.
For 24 hours, Twitter has been under fire from critics, because of the operation of its automatic image cropping system.
This computer code was designed to select a part of a photo that is too large or vertical, in order to enter the cropped subject in horizontal format, and to display it on its site and in its mobile applications.
The story begins last Friday, after a Quebec teacher, Colin Madland, posted a message on Twitter lamenting that the
Zoom
messaging software
wipes the head of one of the members of the university, when he uses the background function. virtual.
“
It turns out that Zoom has a rotten face detection algorithm that wipes out black faces,
” he wrote.
The photo below speaks for itself.
Zoom's virtual background removes the face of a black person.
Twitter screenshot
This is a clear example of algorithmic bias.
Machines are fed hundreds
This article is for subscribers only.
You have 87% left to discover.
Subscribe: 1 € the first month
Can be canceled at any time
Enter your email
Already subscribed?
Log in