Artificial intelligence: Is AI about to rewrite music history?

Had it been released with commercial interests in mind, instead of by a not-for-profit organisation with noble intentions, then the “new” Nirvana song which dropped a few weeks back would have made for a much louder and fiercer talking point.

Drowned in the Sun was no mere forgotten off-cut posthumously unearthed from the archive to mark 27 years since Kurt Cobain’s suicide, but rather an original composition, generated, in a manner of speaking anyway, by the man himself from beyond the grave. How on earth?

Support The Big Issue and our vendors by signing up for a subscription.

Lost Tapes of the 27 Club is an innovative music project by Toronto-based non-profit Over the Bridge that’s all about raising awareness of mental health problems among musicians.

They’ve used the latest AI technology to analyse up to 30 songs each by selected members of the so-called “27 club” – that often distastefully mythologised group of musicians from Cobain to Jimi Hendrix, Jim Morrison and Amy Winehouse, all of whom died at the age of 27 after mental health struggles. Humans have then worked with this AI to co-write and perform “new” (non-official) songs in the artists’ signature styles.

Think of it like rock’n’roll’s CGI Princess Leia in Star Wars moment – not wholly convincing, but unsettlingly impressive nevertheless

Featuring the ragged voice of Eric Hogan, lead singer with “ultimate” Nirvana tribute band Nevermind (AI isn’t yet advanced enough to convincingly copy voices, but give it time), Drowned in the Sun was unveiled in early April. Three minutes of ersatz sludgy-brooding guitars, knock-off Dave Grohl drum fills and neurotic-by-numbers lyrics (“I still got some pain / but it’s over now”), that sounds creepily like the real thing, in a very wonky kind of way. Think of it like rock’n’roll’s CGI Princess Leia in Star Wars moment – not wholly convincing, but unsettlingly impressive, nevertheless.

As well as successfully drawing attention to mental health problems among musicians – the statistics for which are sobering (71 per cent have experienced anxiety or panic attacks, 68 per cent have experienced depression, suicide attempts are twice as common as among the general population) – the Lost Tapes project can’t help but also highlight some increasingly pertinent questions around AI and the music business. Because what is actually stopping this kind of thing being done with commercial interests in mind?

The AI revolution has been coming to music for years now, with dozens of start-ups and special divisions of major corporations around the world – including Spotify, with their faintly sinister-sounding Creator Technology Research Lab – pumping millions into exploring the intersection of music and technology. This June the second AI Song Contest will be staged online – an international competition in the image of Eurovision, originally conceived by Dutch public broadcaster VPRO, which invites the public to vote for their favourite AI-human co-generated pop song (Australia’s Uncanny Valley won last year with a cheesy electro-disco number informed by, among other things, audio samples of koalas, kookaburras and Tasmanian devils).

Should we be worried? We already let AI dictate much of our cultural diet through algorithmically selected songs, movies and TV shows on streaming services – so why not allow it to go deeper? From Irving Berlin to Max Martin, we’ve long been comfortable with celebrating the work of huge behind-the-scenes hit songwriters – why should the lack of a beating heart disqualify machines from joining in? Could AI one day be intelligent enough to listen to all of Max Martin’s music and just do what he does?

Not for a while yet. Clever as AI may be, it’s still only apt to churn out a kind of 21st-century muzak. It remains a long way from learning how to replicate the personality, authenticity and human emotion that defines true songwriting and performance. But that may change, and when it does, the music industry had better be ready – for instance by formulating a new digital-age law that protects artists’ material from mimicry by AI-generated voices. “Intelligence rights” may one day need to be written into musicians’ contracts, otherwise technologically rehashing music’s past for profit may become an industrial-scale free-for-all that knows no bounds. Pop may truly eat itself.

A dystopian, almost ludicrous-sounding proposition perhaps. But technology itself is in many ways the new rock’n’roll for Gen Z, to whom the likes of Apple and TikTok are the epoch-defining innovators and trendsetters that Hendrix or Cobain were to their parents or grandparents. If AI is one day smart enough to create music that sounds as convincing as human songs, will future generations really be fussy about choosing between the two? Or will they simply reason: here we are now, entertain us?