Spotify is a Swedish audio streaming platform that provides digital content from record labels, including Sony, Universal Music Group and Warner Music. The platform is used by millions of users across the world to stream music and has become one of the leading subscription-based music services. However, Spotify is now transitioning to lend its support to advanced machine learning algorithms and AI-driven technology to help the platform curate better music.
The company is reportedly making use of counterfactual machine learning to suggest new music to users. The concept is similar to the movie ‘Sliding Doors’, but with Spotify’s music collections. The team of researchers at Spotify are utilizing a machine-learning model which can capture the complex mathematics behind counterfactual analysis. It can be used to identify the causes of past events and help to predict future ones, such as the user’s next favorite album or playlist.
The concept is inspired by Albert Einstein’s light-cone analysis. In his analysis, he devised the idea of a ‘futures cone’ – the widest part of the cone signifying the ‘preposterous’ or highly unlikely, and the narrowest circle pertaining to the ‘probable’.
Spotify’s machine-learning model is being tested with real-world data, including the impact of water sources in Kenya. However, many are apprehensive about the use of AI-driven technology and how it could possibly impact users. Similarly, major tech giants, such as Meta, LinkedIn, Amazon, and ByteDance are all utilizing causal inference in machine-learning tools for numerous applications.
Spotify is one of a kind service and its efforts in pushing the limits of technology doesn’t go unnoticed. Its current intentions of combining the latest in machine-learning and AI technology to curate playlists and match users with their preferred music style is revolutionary and has been warmly welcomed. Despite this, many would prefer that Spotify uphold their promise with Spotify HiFi and leave out the TikTok-style revamp.
Overall, it will be interesting to witness the results of Spotify’s efforts and the implementation of counterfactual machine learning. While it may provide better personalised music recommendations, it remains to be seen whether or not the technology will actually be beneficial or even an effective substitute to traditional music curation.
