An algorithm for suggesting songs to users on music streaming platforms favors male artists.
This is what researchers say in a study presented in March at a conference in Australia.
This artificial intelligence would thus amplify the gender inequality that already exists in the music sector.
The researchers made this finding after analyzing the habits of 330,000 streaming service users over a period of nine years.
An algorithm to reverse the trend
So, on average, a subscriber has to wait seven or eight tracks before hearing a song performed by a woman.
Only 25% of all songs listened to are sung by a female artist, the study shows elsewhere.
The authors evoke in
The Conversation
a true vicious circle: the algorithm relies on the songs already listened to to generate its suggestions.
Researchers have developed a process to stop the process.
It consists of reorganizing the recommendations generated by the AI.
By moving back a few places in the list of pieces performed by men, the authors ensure that they can gradually offer singers greater visibility.
The use of their modified algorithm in fact results in a greater number of female songs listened to.
Scientists specify, however, that in terms of gender inequality in music, "the responsibility does not rest only on the streaming platforms".
In particular, they point the finger at festival programmers.
High-Tech
Spotify: The prices of certain formulas increase by one or two euros in Europe
High-Tech
Netflix: French subscribers are 41% to share their account outside their home
High-Tech
gender equality
Inequality
Streaming
Women
Song
Man
Music streaming