The Algorithmic Echo Chamber: Is Spotify’s AI Harming Australian Music?
Recent research commissioned by the Victorian Music Development Office, with a specific focus on Spotify, has delved into the complex relationship between artificial intelligence and the Australian music industry. While the notion of AI “killing” local music may be an overstatement, the findings reveal a landscape where algorithms, designed for user engagement, inadvertently create an environment that makes it challenging for emerging Australian artists to gain traction.
This investigation follows a previous report by former Spotify chief economist Will Page for the Australia Institute, which suggested a significant decline in the presence of Australian artists within the top 10,000 most-streamed tracks on platforms like Spotify, YouTube, and Amazon. Page’s analysis highlighted a 30% drop between 2021 and 2024, attributing this to algorithms that, while recognising language, disregard geographical context, thereby failing to recommend local music to Australian audiences. These concerns are echoed by many Australian musicians who feel their work is overshadowed by the global appeal of more established international artists.
Our new research, conducted in February 2025, analysed a substantial dataset of 2.27 million music tracks via Chartmetric’s real-time analytics platform. This comprehensive study encompassed 12,333 artists and 5,000 editorial (human-curated) and AI-generated Spotify playlists from seven English-speaking nations: the United States, the United Kingdom, Australia, New Zealand, Canada, Ireland, and Jamaica.
How AI Shapes Our Listening Habits
The core objective of streaming platforms like Spotify is to maximise user engagement. They achieve this through a multi-pronged approach to music discovery, which includes:
- Manual Search and Exploration: Users actively seeking out specific artists or genres.
- Editorial Playlists: Curated lists created by human editors, often focusing on themes, moods, or new releases.
- AI-Recommended Playlists: Algorithms that suggest music based on a user’s listening history and inferred preferences.
While AI-recommended playlists offer convenience for many users who prefer to be passively served music they might enjoy, they have also drawn criticism for potentially amplifying the influence of superstar artists and the commercial interests backing them. This can lead to a narrowing of listeners’ musical horizons, creating what is often termed a “filter bubble.”
The Algorithmic Bias Unveiled
Our research indicates that AI-generated playlists, particularly for the Australian market, exhibit a strong reliance on global listening patterns. Crucially, these AI-driven selections are less inclined than their human-curated editorial counterparts to feature diverse or regionally specific music. This pattern aligns with observations made in the UK market, suggesting a broader algorithmic tendency.
A significant finding is the way AI recommendations reinforce the dominance of the US music market. By replicating US tastes as global “norms,” AI playlists across all studied countries showed a striking similarity to those originating from the US. Given the sheer size of the US market, its music output inherently dictates global trends, a phenomenon that AI algorithms readily perpetuate.
Furthermore, our analysis revealed that AI playlists draw from a considerably smaller pool of unique tracks compared to editorial playlists – approximately a quarter as many. This concentration further illustrates how AI recommendations are less likely to surface local or niche music.
This tendency for AI to favour “familiar” music disproportionately benefits artists from dominant markets like the US. Our sample data showed that a substantial 77% of US tracks were produced by “established artists” – categorised by Chartmetric as legendary, superstar, or mainstream. In stark contrast, only 22% of Australian tracks fell into these established categories. The remaining 78% of Australian tracks, by less established artists, face a significantly lower probability of being recommended by AI algorithms.
The ‘Rich Get Richer’ Phenomenon
The continuous loop of AI playlists favouring established US artists creates a self-reinforcing cycle. This perpetual exposure to already popular acts further disadvantages emerging talent, solidifying a “rich get richer” dynamic within the streaming ecosystem. These conditions present a formidable barrier for up-and-coming Australian musicians attempting to break through Spotify’s recommendation systems.
While a potential solution could involve Spotify actively tailoring its AI algorithms to boost less-established artists, the intricate workings of these systems remain largely opaque. Until such adjustments are made, the challenges for Australian artists seeking broader algorithmic visibility are likely to persist.
Mohsin Malik receives funding from the Victorian Music Development Office for this research. Guy Morrow also receives funding from the Victorian Music Development Office for this research. He is affiliated with Save Our Arts (Australia) and serves as president of the International Music Business Research Association.
