The record of the year category for the 2025 Grammys is full of zesty pop hits from young female acts such as Chappell Roan, Charli XCX and Sabrina Carpenter.
There’s also Kendrick Lamar’s operatically vicious “Not Like Us” and some poignant, expansive work from Beyoncé (“Texas Hold ’Em”) and Billie Eilish (“Birds of a Feather”).
Then there’s the Beatles’ “Now and Then”. The quartet is back on the Grammy leaderboard a full six decades after winning their first statuette.
“Now and Then”, salvaged from a famously muddy demo from John Lennon, was made possible with the AI-driven, instrument-isolating mix technology first showcased in the documentary series The Beatles: Get Back.
Not even the deaths of Lennon and George Harrison could stand in the way of the most tantalising prospect in rock – a new and final Beatles single, featuring all four members together.
The Recording Academy lauded the single with record and rock performance nominations. The music industry saw the achievements of “Now and Then” as a major feat of production technology and songcraft.
But the academy has also set hard rules around where AI can aid in making music and where it is disqualifying.
“Now and Then” is perhaps the best-case scenario for AI’s place in music.
It’s a pearl of music history, made possible through subtle technology that illuminates, rather than generates. But will its Grammy success open the floodgates for more veteran artists to access and alter old recordings?
“I think AI is a bit like nuclear power. It can split the atom – is that a good idea? Yes if you’re creating energy, but no if it’s a bomb,” said Giles Martin, producer of “Now and Then” and son of long-time Beatles’ producer George Martin.
“For me, when I listen to John’s voice, without fabrication, I felt like I was with him. That’s almost the opposite of AI.”
In 2023, the Recording Academy laid ground rules for how music can incorporate artificial intelligence and still be eligible for awards. The rules say that “only human creators” can win Grammys, and that “the human authorship component of the work submitted must be meaningful”.
Source: scmp.com