Can AI Music Ever Feel Human? The Answer Goes beyond the Sound

Can AI music ever feel human? It's not just about the sound

TO Denis Ellis Bechard edited by Clara Moskowitz

This week I went to Suno, artificial intelligence music platform. I just read new research As a result, it turned out that most of the participants could not distinguish between Suno's music and human works, and I wanted to try it myself. I thought of a song that meant something to me—”For What It's Worth” by Buffalo Springfield. I first heard this tune when I was 17 years old, sitting in my stepfather's kitchen in rural Virginia as he sang and played a guitar he made by hand. Released 30 years ago, in December 1966, the song was a response to the counterculture era curfew riots on the Sunset Strip. clashes between police and young people in Los Angeles. With my guitar in hand, I began to learn the chords, trying to understand feeling it gave me.

Now at the computer I suggested AI create “folk rock” protest song1960s vibe…male vocals with a serious tone.” Generation took seconds.

I put on my headphones and listened, imagining myself in a cafe as the song played on the sound system. Although knowing what it was created by artificial intelligence made me look for signs of artificiality, I doubted that I could distinguish it from a man-made song. And although it didn't give me shiver or make me want to play it on repeat, most songs don't do that.

About supporting science journalism

If you enjoyed this article, please consider supporting our award-winning journalism. subscription. By purchasing a subscription, you help ensure a future of influential stories about the discoveries and ideas shaping our world today.

The paper on AI music, a preprint that has not yet been peer-reviewed, is based on thousands of songs on a Reddit board where users post music created by Suno. The researchers then presented study participants with pairs of songs and asked them to identify which of the tunes was created by artificial intelligence. The team found that participants made the correct choice 53 percent of the time (close to guessing), although when presented with stylistically similar songs from humans and artificial intelligence, their accuracy was as high as 66 percent. But the AI ​​generation models update frequentlyand by the time the study was released as a preprint, more advanced Suno model was available.

Our relationship with the music has changed keeping up with technology. IN 2002 interviewDavid Bowie thought that everything in music would soon change. He predicted the transformation of its distribution and the disappearance of copyright. And to highlight how easy access would be, he said, “The music itself will become like running water or electricity.” This could hardly be called a prophecy. Napster, a music sharing platform launched in 1999, has already opened up electronic music sharing and piracy, making music distribution easier than ever. Then, in 2003The iTunes store started selling songs for 99 cents apiece, and in 2008 Spotify monthly subscription now available taps are even wider. Since 2023, Suno has been contributing to the increase in the amount of music distributed online. Spotify recently announced that the company has removed 75 million “spammy” music tracks over the past year to maintain the quality of its offerings, although it is unknown how many of the removed tracks were created using Suno.

But even as AI music improves, I can't help but wonder how it will fit into our lives. I grew up on mixtapes, the precursors playlistsmade for training, traveling or just sharing. They were made up of songs that I or my friends liked. I can't imagine someone handing me a flash drive of AI-generated tracks and saying, “There are infinitely more where they came from!”

However, history teaches us not to underestimate how a few inventive people can use new technologies to express themselves. In the 1970s, as disco DJs expanded and edited songs for the dance floor, remix culture was born. In the 1980s, hip-hop artists selected funk, soul and rock songs to create new tracks. I grew up hearing people call sampling lazy and compare it to stealing; federal judge reopens 1991 case indicative sample conclusion with the biblical quote “Thou shalt not steal.” When Danger Mouse combined Jay-Z and the Beatles into Gray album in 2004, music label EMI sent out cease-and-desist letters. Fans, however, staged a “Gray Tuesday” to spread the mashup. in protest. Now we recognize the art in editing, and the DJs have moved from the corner to the tent.

But DJs have always had something to do with their music – they sample songs they like. Although one could argue that AI (which also has copyright problems) trained in music that people love, it's not easy for us to feel it connection; The most we can say is that in some sense it has a terroir.

It's hard for me to see how AI music will beat us. TO 2015 And 2017Research has already shown that people cannot distinguish human-produced music from computer-generated music. And a few years earlier, in 1997, composer and computer pioneer David Cope created music software. The audience who heard the pianist perform his piece along with a piece by Johann Sebastian Bach thought that the software composition was real Bach.

So while there is reason to fear that talented musicians will never be heard as millions of people who don't play instruments or sing flood the Internet with AI songs, I suspect that most AI music, like most other music, will be forgotten or never seen. Even in exceptional human music, we need more than just virtuosity—an origin story, a connection. Likewise, a few rare, unusual AI songs will no doubt be tied to cultural moments—movies, videos, memes—or created by AI music studios, which will give people more control over the outcome than text prompts, and that could allow for more innovative and personal songs.

This doesn't mean we shouldn't be careful with music machines. A decade after mass production of pianos and phonographs began in the 1890s, the composer John Philip Sousa warned that amateur musicians will disappear and people will become “human phonographs.” The fear was not in vain. Historically, many families played music together, and this tradition has faded. Whereas parents and children used to play in the living room, now teenagers sit alone in their bedrooms and play songs on headphones. By comparing music to electricity, Bowie was referring to a similar loss. He said that musicians should be prepared to tour frequently, implying that live performance will be the only way to create a genuine connection with the audience – “that's really the only unique situation that will remain,” he added.

When my stepfather taught me to play “For What It's Worth,” he shared a song he first heard on the radio when he was nine years old, and that at age 11 he learned to play by listening to vinyl record he bought it for a couple of dollars. I hear this song often now in cafes and at yoga classes – it's back in fashion, as relevant to today's social issues as it was in 1966 – and I notice it every time, not because of its melody, lyrics or virtuosity, but because his story.