ChatGPT or handcrafted work? The case of the 500-album playlist
Adding 500 albums to a pogre metal playlist takes four hours, according to a detailed analysis. But AI tools can generate similar lists in seconds. The difference between handcrafted work and a well-trained prompt is no longer obvious, and a recent case has put the spotlight on the authenticity of playlists.
How the accusation works
The suspicion stems from the structure of the playlist: after analysing it, patterns typical of algorithmic recommendation were observed, such as the grouping of very specific niche bands and the appearance of freshly released albums in positions that do not match a random manual selection. The accused claims he added each album one by one over four hours, using the Trackify program to keep track of his listening. However, sceptics point out that certain details, such as Akercocke appearing at the top of the list when it should have been at the end, are hard to explain without algorithmic intervention.
The evidence in the playlist
The playlist in question contains 500 albums, mostly pogre metal, atmospheric black metal and avant-garde. It includes names such as Opeth, Ulver, Akercocke, Yellow Eyes, Messa and Hiatus Kaiyote. An analysis of a daily top 8 showed Bill Frisell leading with 129 plays, trinc by Yellow Eyes with 118. These numbers, according to critics, reflect a listening profile too consistent with the recommendations of an AI assistant. The author of the analysis himself admitted that the chatbot told him Panopticon was one of the albums of the year, which for some is an implicit confession.
Trackify and prompts: the technical side
Trackify is a program that lets you track the most played albums on Spotify, but the user complained that it did not keep a complete history, forcing him to rebuild the list manually. He later corrected the mistake and confirmed it did work. To generate the playlists, he also used Playlistable, a tool that can integrate with AI. The key question is whether a well-designed prompt can imitate personal taste to the point of fooling other music lovers. Defenders of the manual approach argue that the diversity of genres and the presence of obscure bands like Kaatayra or Hail Spirit Noir are difficult for a general-purpose algorithm to replicate.
The final question is not whether ChatGPT was used, but when it will stop mattering. If a list sounds like an algorithm, is it less authentic? The debate remains open, and the next time you share your playlist, someone might just ask you for the prompt.
Related discussions in the forum