With the success of Peter Jackson’s Get Back, the documentary streaming on Disney Plus, Beatlemania is back. Watching Paul McCartney create the eponymous song out of seemingly nothing, as George Harrison stands nearby yawning, is one of 2021’s cinematic pleasures.
The Beatles are arguably the most successful pop group in history and, in the years since their heyday, countless artists, producers and songwriters, not to mention record companies and now music streaming services, have tried to recreate the same magic. The latest tool for capturing elusive pop music gold is artificial intelligence.
Usually when we think of artificial intelligence creating art, it’s making something bizarre or unintentionally hilarious. Take Google’s horrifying Deep Dream with its thousands of dog eyes, or Sunspring, a movie written by an AI that was fed hundreds of sci-fi scripts. It’s not just movies that have been influenced by machines but music too, with brand new AI-written songs designed to sound exactly like Nirvana and Jimi Hendrix.
The use of AI in music is here today, and while its contribution is still small, it will likely grow. If you take that to its logical conclusion, the question arises: Could AI create pop stars — and Grammy award winners — completely out of thin air?
The digital pop star
It’s easy to imagine the day when artificial intelligence will be used to the same extent that sampling and home recording are today. The humble computer already enables musicians like the Grammy-winning Billie Eilish to record, sample and sequence songs at home. It’s cheaper than recording in a studio, which helps cash-strapped musicians and maximizes profits for record companies.
AI tools can already create entirely new music from scratch, from apps that produce automated lyrics to others that write the chords and instrumentation. Just as computers and mobile phones have democratized the distribution and recording of music, its proponents say that AI makes the process of music creation available to everyone. It could also have some unforeseen negative effects, according to its detractors.
In the TV show Alter Ego human singers use digital avatars to compete in a singing competition. From there, it’s not so far a leap to imagine an artist who’s never physically existed at all. A few AI pop stars have emerged in recent years, including Yona and Lil Miquela, but neither of them are household names. And human artists have been using AI tools for years, including the band Yacht, which used them for the album I Thought the Future Would Be Cooler.
Taryn Southern is a musician and filmmaker who uses artificial intelligence in her art, and the BBC even wondered if she could have been the world’s first number one AI popstar. In an interview, Southern told CNET that AI has the potential to revolutionize music-making in the same way that YouTube revolutionized video production.
“It’s putting very cheap tools in the hands of millions of people, making music more accessible,” Southern said. “I’d read a couple of articles about AI, and so it sparked my curiosity. I thought: Why not play around with some of the tools and see what all this is about?”
She was inspired to experiment with a program called Amper to create her 2018 record I Am AI. While the program wrote the music, Southern still contributed the lyrics and melodies. She’s wary of a completely AI-generated artist.
“I think people like listening to music that they believe has some sort of emotional basis, or if there’s some human emotional element that’s creatively pulling the strings”, Southern said.
AI is already among us
Software such as Alysia and Orb Composer enable users to create songs based on styles or “feel” and are the next step towards a wholly computer-generated pop music future. Amper, now owned by Shutterstock, is another tool available to the budding musician. Co-founder and former CEO of Amper Music, Drew Silverstein, says the program he helped create isn’t much of a leap from existing systems, even if they aren’t labeled as “AI”.
“I mean it already exists. If you go into GarageBand, like, there’s auto drummers that you can set up already,” Silverstein said. He considers AI is a natural evolution of music-making technologies that seem “both amazing and scary at the same time”.
“You know, in a year or 100 years AI music will be old hat. There’ll be a period of time or period in time where someone looks back and says, ‘Ah, that was a really novel thing they did, those rudimentary people,'” he said.
Silverstein says that a hit song completely written by an AI could happen within “a couple of years, if not sooner,” and that an AI instrumental created today would have a decent shot at being at the top of charts. In the end though, he says that AI will help artists in ways never before dreamed of. “I think it’s a fair statement to say that AI music will lead to the greatest creative revolution of all time.”
But if someone creates a song out of an AI, who owns the copyright? Silverstein says it depends on the artists’ input, but inevitably a song written entirely by an AI would credit the programmers.
Pop songs are increasingly recorded at home and often written to a specific formula, as journalist John Brook explains in the New Yorker. For a song to become an “ear worm,” that sticks in people’s heads, it ideally has a hook every 7 seconds and a runtime under 3 minutes.
Silverstein said that hooks aren’t a problem — AI can write them now– but that not even seasoned music industry insiders can predict what’s going to resonate with listeners. He said that an AI needs to be collaborative, so that the person using the tools can make artistic changes such as tweaking hooks.
Artists have recycled hooks from older songs for years. Flo Rida’s Right Round samples Dead or Alive, and Fatboy Slim made a career out of it, starting with Dub Be Good to Me in 1990. There are even brand new songs by popular albeit long-broken-up bands. In 1995 the first new Beatles song in 26 years, Free As A Bird, was released. It used a 1977 cassette recording of John Lennon, and the surviving members built a song around it. As the popularity of Get Back demonstrates, the public is still hungry for new Beatles material.
In 2015 the scientists at Sony’s CSL Labs decided to try their hand at creating their own Beatles tune. Daddy’s Car was written “in the style of The Beatles” by an AI that was fed a bunch of Lennon-McCartney songs. The lyrics were suggested by Beatles song titles, and the resulting song has the hazy sound of the group’s psychedelic period. It may lack the basic verse-chorus structures you’d expect from a pop song, but the result hints at what might be possible.
In 2021, an organization called Under The Bridge — which deals with mental health in the music industry — created new songs in the style of the so-called “27 club”: Nirvana’s Kurt Cobain, The Doors’ Jim Morrison and Jimi Hendrix. At the moment these songs are sung by impersonators, but one day AI will enable artists’ voices to be synthesized.
This effort followed the creation of “new” songs by Katy Perry and Elvis Presley from another research lab, OpenAI, for its Jukebox project in 2020. However as CNET’s Amanda Kooser noted at the time: “Jukebox hasn’t mastered the art of the hook.”
These AI music creations may have seemed far-fetched once, but they’re not so far from those concerts performed by holograms of Whitney Houston and Amy Winehouse.
What effect will AI have on the music industry?
Steven Hawking and Elon Musk alike have warned that artificial intelligence is the “biggest risk we face as a civilization.” It’s true that robots have a bad rap, thanks to TV shows like Black Mirror, yet many companies, even Musk’s Tesla, use AI. Google and Amazon even use the term “machine learning” to give it a less menacing connotation. But the flipside of AI’s “democratization”, as with any new technology, is that it tends to upend the status quo, for good or bad.
While record companies are likely excited about the low costs of using AI, and record stores continue to enjoy a vinyl-led renaissance, it’s the recording studio that is most at risk. Abbey Road, where the Beatles recorded their album of the same name, is arguably the most famous studio in the world. It has persevered despite multiple threats including the pandemic, but surviving AI could be its biggest challenge.
Legendary engineer and musician Steve Albini runs his own recording studios in Chicago, but cut his teeth at Abbey Road in the late 1980s when samplers and drum machines were first having an impact. “Places like Abbey Road were actually sort of falling fallow, and so they started to lower their rates to try to attract more budget-conscious clients,” he told CNET.
Albini says while he believes his own studio can weather any changes, he is suspicious about the future of large facilities in a wholly digital future.
“I honestly don’t know that it’s going to be possible to operate a big professional institutional studio going forward,” Albini says. “I can’t fault someone saying to his bandmates ‘Why pay money to go into a studio when I can do it at home for free?'”
CNET alum Steve Guttenberg, also known as the Audiophiliac, has worked on recordings for New York label Chesky Records over the course of many years. When asked how recording studios like Abbey Road survive in the future, he replied: “They can’t”.
The folks who work at Abbey Road have a more optimistic take. Mirek Stiles, head of audio products, says the industry has always experienced change, and part of his job is about staying abreast of trends. Stiles sees a future where Abbey Road can co-exist with both home recordings and the inevitable rise of AI songwriting.
“In the ’50s music was chopped up and edited and composers would say ‘That’s not my music.’ Then in the ’80s MIDI came in, sampling, and there were actual lawsuits”, Stiles said. “Who could have predicted streaming 20 years ago? In 20 years there could be something else.”
Musician Andy Falkous has had a tumultuous relationship with record companies, but has also used Abbey Road’s mastering facilities. “I think decent music surrendered to the music industry a long time ago,” he said. “I don’t care. All you’ll be left with are the superstars, and then the independent bands. Egregiously independent.”
He cryptically added that the difference between pop and rock is just like the difference between cheese and onion.
“I couldn’t give a burnt fuck about the artificial cheese industry, either. I’ve had a lot of substandard onions and I’m a big fan of the onion,” Falkous says.
The only constant is change
After 50 years, no artist has managed to match the success of the Beatles, and yet as an entity “pop stars” still endure. The music industry is constantly evolving due to the influence of new technologies such as streaming, sampling and auto-tune. As AI music evolves as well, it will have a monumental impact on music production. Meanwhile the old guard — recording studios and artists — are having to adapt.
Every generation throughout history has been told “that’s not the way you make music.” Just as it was for McCartney, strumming a bass like a guitar to write Get Back, so it will be for the next generation of hit-makers. They’ll use AI to unlock brand new sounds, startling and heretofore unknown. And one of them will definitely win a Grammy.