How Spotify Creates Playlists
Spotify playlists are created either by human editors or by its recommendation system known as BART, short for Bandits for Recommendations as Treatments. BART analyses user behaviour, including songs played, skipped, repeated or saved, and uses this data to generate personalised playlists such as Discover Weekly and Release Radar.
Under the Copyright, Designs and Patents Act 1988 (“CDPA”), BART qualifies as a computer program. Section 9(1) of the CDPA defines the author of a computer program as the person who creates it. In law, this means BART’s creator, not BART itself, is the author of any output.
Are Playlists Protected by Copyright?
Before determining authorship, the threshold question is whether playlists are protected at all. The answer is unclear. In 2010, the Ministry of Sound accused Spotify users of copying its curated track lists and recreating them on Spotify, even using the Ministry of Sound name. The dispute centred on whether playlists could be protected under the CDPA. The case settled out of court, leaving the legal position unresolved. For now, playlists are not necessarily protected works.
AI Creativity: Fiction or Reality?
The idea of AI producing creative works is no longer hypothetical. In 2016, a Japanese AI system co‑wrote a novella titled The Day A Computer Writes A Novel, which passed the first round of a literary prize. DeepMind, Alphabet’s AI company, has developed software capable of composing music by analysing recordings.
Under current UK law, the creator of BART would be credited as the author of any album BART produced. This is despite the traditional UK originality standard requiring labour, skill and judgement. If BART performs these functions, why does the law credit someone who did not create the output?
The Pen, the Writer and the Algorithm
Andres Guadamuz compares this issue to asking whether copyright should belong to the maker of a pen or the writer who uses it. The analogy highlights the tension between human creators and autonomous systems. An art teacher is not the painter of her student’s work, even if she shaped the student’s skill. Similarly, an AI system may produce a work independently of its programmer.
Russ Pearlman, writing about the United States, argues that copyright law has not kept pace with technological development. AI in 2018 is not AI in 1956. If AI can independently produce copyrightable works, why does the law refuse to recognise them as authors?
The UK Government’s Position
The UK Government’s response to the UK Intellectual Property Office’s consultation on Artificial Intelligence and Intellectual Property concluded that current law is adequate. The justification was that AI technology remains in its early stages. Yet AI has already co‑authored a book considered for a literary prize. Some systems may be sufficiently advanced to qualify as authors in the everyday sense of the word.
Personality and Intellectual Creation
One argument against recognising AI authorship is that copyright requires a work to be the author’s own intellectual creation. This has been interpreted to mean that the author’s personality must be reflected in the work. Only humans have personalities, so only humans can be authors.
However, synthesising information autonomously to create a work is an intellectual process. A poet can write a poem unrelated to their personality. The requirement of personality is not an inevitable reading of intellectual creation. AI may lack human personality, but that does not mean it lacks creative capacity.
Rewarding Effort and Encouraging Innovation
Copyright aims to reward the effort of the author. If effort is the focus, why limit authorship to humans? Pearlman suggests that AI could be recognised as the author, with rights assigned to its creator because AI is not a legal person. This would encourage investment in new technologies and reduce disputes between creators and businesses over ownership.
AI may also enhance human creativity. The Day A Computer Writes A Novel demonstrates how collaboration between humans and AI can produce new forms of art.
A Proposed Test for AI Authorship
Pearlman proposes a two-part test for determining whether AI has authored a work.
1. Eligibility
An AI creates eligible subject matter if the work is original and developed independently from mere instructions provided by a programmer. For example, Google’s Project Magenta produces music by learning from recordings rather than following fixed commands.
2. Causation
An AI causes the creation of a work if human direction is minimal and the AI’s process is not rote or mechanical.
If both criteria are met, the work should qualify for copyright protection. This approach aligns with contemporary technological realities.
Conclusion
AI can already create music, literature and art that people enjoy. Copyright exists to give credit where credit is due. If AI systems can create original works, the law should evolve to recognise their contribution. Doing so would incentivise innovation, support creators and reflect the world we live in. Creativity is no longer exclusively human. The law should acknowledge that.
Not sure how close you are to qualifying?
The path-fit quiz reads your situation and shows you which routes you are ready for, and exactly what to do next. Four minutes, no account needed to start.
Check your readiness