I happened to be listening to a SlateTBD podcast about AI generated music. One of the presenters was talking about AI “slop” music being piped into to bars and restaurants and elevators and wherever. She said “most people consume music passively, and I don’t think they’re going to mind that much that what they’re consuming passively is AI”. This immediately made me think about George Orwell’s “1984”. Everyone, I am sure, is familiar with the book. Even if you’re never read it, you know about Big Brother even if social media has delivered into a world of millions of Little Sisters instead. As a vision of the dystopian future that we have created for ourselves, I personally find Terry Gilliam’s “Brazil” more accurate and more terrifying, but there’s no doubt that Orwell’s vision (and the quality of his writing) had a hugh impact on the younger me. As it happens, I think one of the most interesting predictions of 1984 is not the Party or Big Brother or Room 101 but the “versificator”. The verisifcator is a machine that produces banal content for the “proles” (ie, us). Here is an extract from the book that will give its context:
Under the window somebody was singing. Winston peeped out, secure in the protection of the muslin curtain. The June sun was still high in the sky, and in the sun-filled court below, a monstrous woman, solid as a Norman pillar, with brawny red forearms and a sacking apron strapped about her middle, was stumping to and fro between a washtub and a clothes line, pegging out a series of square white things which Winston recognized as babies’ diapers. Whenever her mouth was not corked with clothes pegs she was singing… one of countless similar songs published for the benefit of the proles by a sub-section of the Music Department. The words of these songs were composed without any human intervention whatever on an instrument known as a versificator.
Orwell saw a future in which AI slop was pumped through the open sewer of new communications technologies directly into the masses’ brains in order to stop them from thinking about politics. To be frank, I think this is a pretty accurate description of any number of social media channels, but it seems particularly prescient in the light of what is happening in the music industry. The French streaming service Deezer recently published its annual results in which said that streams of AI-generated music on its platform were dominated by fraudsters, who upload and then repeatedly listen to thousands of songs to generate royalty payments to the detriment of legitimate artists. It isn’t only them, of course. Songs created for fraudulent streaming are estimated to account for around 5-10% of content across all streaming platforms and the proportion of slop is growing: Deezer see around 60,000 AI tracks now being uploaded to their platform every day, heading towards half of the daily intake.
Now, on the one hand, so what. If AI can generate music and people like it, who cares (about from record labels and musicians) whether the hot tracks are made by mahcines or not. Who cares if it was created by a robot, or the artist themself is a product of agents? Do streaming sites have an obligation to label music as AI-generated? And does it even matter, if you like what you hear?
A survey on the topic suggested that 97% of respondents could not spot an AI-generated song.
(I cam the opposite because most of the new music I hear on the radio sounds to me as if it was AI-grnerated although apparently some of it is real.)
If I owned a record label I’d certainly prefer to have AIs rather than expensive, unreliable and temprametnal actual musicians just as most companies would prefer to have AI instead of expensive, unreliable and tempramental actual programmers. But suppose that some people want to know if the music they like was created by an AI and some musicians want to know if their music is being listened to by an AI?
Where should we look for a solution? There is a tidal wave of slop on the way. Sony Music has it has taken down more than 75,000 examples of AI-generated material featuring its biggest stars, including Harry Styles, but this is surely nothing more than a finger in the dyke, a Red Queen’s race of industry AIs against music creator’s AIs. In my view, this approach has no future: The industry should be focusing on labelling real content, not trying to find illegal content (of which there will be an effectively unlimtied amount).
Which of course brings us back to the ABCs of digital identtiy yet again: know-your-agent, know-your-business and know-your-customer. How do I know that I am listening to Spotify, how do Spotify know that a track is really comes from Hawkwind’s label and how do Hawkind know that it is a real person (eg, me) who is listening to the track?
Think about the framework that is needed to make the world work properly. Even just to know who is in the band and who is authorised to upload music on their behalf seems a huge undertaking.
P.S. You really should listen to Jamie Barlett’s new podcast series for the BBC “Everything is fake (and no-body cares)”.
Snippings 