Once upon a time, in a land far, far away, there was a pop group with five members, who could all sing, dance, create viral challenges on social media and personally answer every message their fans from all around the world sent to them, in any language.
They would also record special video messages for each of their fans, taking an interest in their personal lives and thanking them for their support. The fans were also given special presents as a badge they could wear to let their friends know how much of a fan they were.
The band released irresistible songs which their millions and millions of fans adored, and they released a LOT of these songs all of the time. They also created very very special live performances where their fans could talk to each other and influence the show as it happened.
The band were so popular for so long, that every now and then one of the members would be replaced by someone completely new, just in case people started to decide that maybe they liked different things. And, just like that, the band somehow became even more popular.
In fact, they were so popular that the people around them forgot to find any other bands or singers that people might like. Until there were only a few bands like them making any new music any more.
Very suddenly, it seemed like all of those bands were making music that was almost the same! People became very very sad because they couldn’t get excited about music any more. And that was a real shame.
___________________
Now, what you’ve just read could be a fairy tale. But does the lack of a feel-good, wholesome family-friendly ending lead you down the road of a nightmarish dystopian possibility?
Virtual artists aren’t new, but now with more sophisticated AI in play, what does that look like? Outlined above, each ‘band member’ handling hundreds, thousands or even millions of interactions and transactions in real time. Bot stars, if you will.
The history of music being released and promoted to the public without the traditional presence of an artist playing live shows or hitting the promotional trail stretches back far into the last century. From novelty bands releasing music from TV shows in the 1960s, through John and Yoko’s Bagism to Gorillaz in more recent times, these artists have all had actual humans involved in the creation, recording and (sometimes) promotion of their music.
The definition of ‘virtual artist’ as we understand it in 2023 has its origins in Japan and South Korea in the 1990s. Early examples used the technology of the time to create avatars to visually represent an entirely fabricated ‘artist’ however, again, the actual music itself was created by relatively traditional means, often with a voice actor taking care of the singing itself.
Here and now, we’re beginning to see the emergence of powerful new technology which can credibly replicate the singing voice, production style, songwriting and visual appearance of established musicians and performers.
Hybe recently launched Midnatt, an AI-assisted project that allowed them to release multi-language variations of the debut track. Tellingly, Hybe chairman Bang Si-Hyuk recently opined to Billboard that he has “long doubted that the entities that create and produce music will remain human. I don’t know how long human artists can be the only ones to satisfy human needs and human tastes.”
Sony Music Entertainment Japan’s Hatsune Miku was one of the first, more advanced notions of what a virtual artist could look like. Having released multiple albums most years since the mid-2010s, they are already a global phenomenon having supported Lady Gaga on tour and currently with over 1.3M monthly listeners on Spotify as a lead artist. Not bad for a glorified vocal synthesizer.
Lucy was revealed by Tencent Music in December 2022, their “first hyper-real virtual pop idol”. The launch was met with a deluge of views on Weibo, extensive media coverage and approaches for brand collaborations.
Apoki is a K-Pop artist with more than 4M social media followers. There hasn’t been a great deal of music so far, but there is a merch line and of course, an NFT offering and a dedicated metaverse (it is 2023, after all).
The majors and others in English-speaking markets have all dipped their toes to varying extents, signing virtual artists, establishing web3/AI imprints, and collaborating with or investing in AI start-ups. But heed the cautionary tale of FN Meka, the virtual artist whose signing was announced in a blitz of publicity before being swiftly dropped by Capitol Music Group amid claims of racism and cultural appropriation. Them pesky humans did a stupid again.
___________________
This toe-dipping by the biggest players in our industry is worth paying attention to. Amidst the fevered panic from the loudest shouters about music created by generative artificial intelligence, there has been a rash of deals with and ‘strategic investments’ in selected artificial intelligence start-ups.
Witness the change in rhetoric following some surprising plot twists in the last few weeks; and how the music industry eventually embraces (and then probably defaults to) this technology.
The more cynical might argue that the music industry has forever been an exploitative and murky world where shadowy overlords pull the strings to obtain and assert their own personal power and wealth. That gifted musicians are chewed up and spat out in servitude of the borderline sociopathic self-aggrandisement of a handful of senior executives. That only an incredibly lucky few artists ever make it anywhere near payment parity with the bloated compensation packages of those top executives.
We all know the Hunter S. Thompson quote.
So, in a world with AI-generated songs, AI-assisted record production, and AI-created ‘virtual bands’, record companies can dispense with all the time and resource spent on the hard stuff (not that hard stuff) - finding, developing, marketing, promoting and breaking new (human) musical talent.
Surely this is an exciting development for the multinational conglomerates - who are ultimately answerable only to their shareholders, despite regular proclamations of being the protectors and promoters of creativity, expression and culture.
Presuming that the technology will develop enough to be able to deliver something credibly close to what we love about music from humans, AND that the factory-floor approach doesn’t fragment us all even further due to the sheer weight of music being released on DSPs, AND somehow as fans we can get excited about an artist that technically doesn’t exist, why then would major music companies continue to spend most of their revenue in new human talent when the virtual equivalent would take way less time and development investment, and be statistically more likely to break through?
With virtual artists racking up hundreds of millions of streams (and very well followed social media accounts) already, this is a headwind that will transform our industry and almost certainly, completely change the furniture within major labels.
The business is only starting to recover its real-terms value following the last technological shock to hit it twenty years ago. Perhaps a vestige of today’s industry will survive - a handful of global (human) superstars, with maybe a few more at domestic / regional level to ensure local dominance, but beyond that? If we’re still in the streaming model we’re in today - a numbers game based on volume and share of listenership - it will be far cheaper and a lot less risky to mine the data, create a virtual artist based on the success of those past, maybe partner with a brand and off you go. Cranking out innumerable tracks by literally faceless acts to flood DSPs with volume to cling on to marketshare.
___________________
Living without the significant cost base of recording, writing, touring, promotion and marketing has its obvious advantages to rightsholders if they can hit the magic formula.
Many are sitting on humungous data sets (built on and by their signed and distributed artists), which could be used to train AI on many different prompt responses to decide what music will work best, where and for which kind of virtual artist.
The irony here being that if those signed and distributed artists have not been consulted on their data being used for large machine learning, it could start to sound similar to a letter we all read about in April this year…shouldn’t artists expect to exercise the same control on their output, metadata and performance metrics? We could already be on ethically and morally sketchy ground, if not legally.
The rise of generative AI’s use in music is already impacting on the career opportunities and businesses of human artists, as well as rightsholders. The gaming of streaming in its current form is denying many real artists the chance to make a living from music. That’s only going to get harder in a world of knock-off Adeles and bootleg Bunnys.
Comments