CAN ROBOTS BUILD BETTER HITS?

Music-industry insiders may dismiss it, but AI songwriting is a trend that’s here to stay

Attention, Ed Sheeran: Artificial intelligence is coming for you. While the idea of a computer-crafted chart-topper seems far off, AI songwriting is gaining traction. Startups such as Amper, Popgun, Jukedeck and Amadeus Code have raised millions of venture-capital dollars on the bet that machines will become valuable creative assistants to artists in the near future. Several artists utilizing a songwriting algorithm called Flow Machines already have appeared on Spotify’s New Music Friday playlists.

How does it work? An algorithm ingests thousands of songs in a specific genre and rapidly cranks out chord progressions and melodies optimized for that style — even indistinguishable (according to another algorithm, that is) from music written by people.

While these programs might not be consciously driving or tapping into musical trends the way human artists do, they churn out content much more efficiently than, say, an exhausted musician who’s been touring for months. This productivity could have obvious appeal to record labels and streaming services: Generate tons of new music without ever having to pay a human writer? Yes, please. Spotify, in fact, drew criticism in 2017 for featuring on several mood playlists artists who did not appear to exist. The assumption: The company was using AI-crafted songs and crediting them to made-up people. But few on the AI development side believe that their creations will replace artists altogether. Instead, they see their algorithms as a supplement to human efforts, offering new arrangements of notes and tunes that can help songwriters clarify their own ideas — enhancing, not neutering, their creative power. Citing electric guitars and drum machines, Amadeus Code founder Taishi Fukuyama says, “History teaches us that emerging technology in music leads to an explosion of art.” CHERIE HU