Spotify’s “new” feature

I recently discovered Spotify’s AI-powered DJ feature for the first time, though it first came out a little over two months ago in late February. The feature provides users with a personalized listening experience by analyzing a user’s listening history, favorite genres, and other preferences to create custom playlists and recommendations. It essentially acts as a virtual DJ, curating a unique playlist for each user based on their individual music tastes. To create the voice model for the DJ, Spotify worked with their Head of Cultural Partnerships, Xavier “X” Jernigan.

At first, I was skeptical.  Since generative AI is a hot topic right now, it seems as if every social media app is implementing some kind of AI-powered feature for the sake of doing it. However, after listening for a while, I started to enjoy it. As an avid music listener, I am always listening to music, but I find that I get sick of songs pretty quick. I was hoping the DJ could serve as another way to discover songs.   

During my first session with the DJ here was what I noticed:

• Good: 
  • DJ switches up the genre, so you’re not listening to the same type of music for a long time. You even have the option to manually switch the genre on your own. 
  • I love X’s voice
• Bad:
  • After a while, the script for how the DJ behaves is pretty rigid
  • I wish I could save certain segments when using the feature

The concerns

Around the same time, I read an article in the Wall Street Journal titled “Why It Matters Whether a Robot Is Given a Gender.” The article discusses how the Brookings Institution and other organizations have raised worries that assigning gendered names, voices, and appearances to technology could perpetuate harmful gender stereotypes. For instance, digital assistants such as Alexa and Siri, which were initially given feminine names and voices, faced criticism for perpetuating the stereotype of women being subservient and meek. 

However, a recent study published in the Journal of Experimental Social Psychology has revealed a potential benefit to assigning a gender to technology: it can increase people’s emotional attachment to it. Regardless of whether the robot is perceived as male or female, people tend to view it as more human-like when it has a gender, resulting in stronger feelings of attachment to the product. As a result, people are more likely to rate gendered technology positively and purchase it.

This poses a dilemma, which the authors address:

“Gendering technology reinforces problematic stereotypes, but it also facilitates anthropomorphism (humanization), with beneficial consequences for the marketing of various technologies.”

According to Ashley Martin, an associate professor of organizational behavior at Stanford University and one of the authors of the study, the issue is not whether stereotypical male or female traits are inherently good or bad. The problem lies in the fact that traits typically associated with women, such as kindness and warmth, are not usually associated with leadership and power, whereas traits often attributed to men, such as decisiveness and independence, are. 

To conclude…

I love the AI DJ. Hearing X’s voice every time I open the AI DJ brings me so much joy, but it feels weird knowing that my affinity for the new feature is rooted in some kind of subconscious gender bias. 

After using the DJ for a while now, I’ve noticed it’s not that revolutionary. It’s a neat feature, but won’t replace how I listen to music. What is more important is the role it plays in the broader conversation of  gender bias and AI. 

For example, since gender stereotypes have been found to contribute to women’s underrepresentation in leadership and management roles technology that reinforces such stereotypes, therefore, could be problematic. However, separating gender from robots will be a challenging task, especially if the goal is to create an emotional attachment in users.

Link to original article published on 5/02/2023