Skip to playerSkip to main content
  • 1 day ago
Transcript
00:00Take a listen to these three snippets. I want you to guess which songs are produced by humans
00:05and which are produced by artificial intelligence.
00:20Well, I tricked you because all of these songs were created with AI, including the last one,
00:26which just became the first ever AI-generated song to hit number one on Billboard's country digital
00:33song sales chart. I know, I know, but that was the point. Most people can no longer tell the
00:40difference between a human-made song and one made by AI.
00:46Ipsos and music streaming app Deezer surveyed 9,000 people across eight countries playing one
00:52human song and one AI song. 97% of respondents got it wrong. So how hard is it really? Kato on the
01:01track is a number one Billboard charting music producer and says there's usually one big giveaway,
01:07but it's something trained ears pick up faster. AI-generated songs still to me
01:13have this very sanitized quality to it where like the singing is perfect. The production sounds like
01:22very clean and like very, uh, like the chord progressions are like so clean. So to test this
01:30out, I brought in the big guns, AKA my sister and music lover and Harry Kumar, co-founder and chief
01:37creative officer of the quantum technology company MOF. Oh, this is gonna be hard. In spring mornings,
01:45the sun rises on the horizon. I think that's AI. I'm inclined to say it's AI. It has like a weird
01:53like metallic sound, but it's also way too on beat for that style of song. What she's hearing,
02:02that too perfect feel is the same thing many producers listen for in music software. Everything
02:09sits on a grid like a musical spreadsheet. When you click notes with a mouse, they snap perfectly
02:15onto that grid. But when a real person plays a melody, even the best musicians are slightly off the
02:21grid, not off beat, just human. Those micro imperfections are what make music feel authentic
02:27while AI hits every note with mathematical precision. I think it's real. I think
02:42potentially the vocals were already sampled. My first instinct was it's not AI.
02:48Oh no, I should have went with what I thought. This second song is all over TikTok. It's now been
02:55deleted, but people have suspected it was AI for some time now. And those rumors grew when it vanished
03:02from Spotify. Back in September, Spotify announced new AI protections, including a ban on impersonation
03:09and AI cloning of artists' voices, unless the artist gives permission. A Spotify spokesperson confirmed the
03:15reason for the removal of I Run to Straight Arrow News, saying this song was removed for violating
03:21Spotify's impersonation policy. No royalties were paid out on the track's streams. To be fair,
03:27our contestants did pretty well overall. But what you saw here is exactly what's happening globally.
03:34AI tools like Suno, recently valued at nearly $2.5 billion, are evolving so quickly some songs
03:41sound indistinguishable from real humans. As I mentioned, Kumar is co-founder and CCO of Moth.
03:47A few months ago, Straight Arrow News reported on Moth's Breakthrough,
03:51the first song ever created using quantum AI music technology.
04:07We used some classical machine learning techniques and some quantum machine learning techniques to then
04:14produce this generative stream of music, where each time you listened to it, it was stylistically
04:20the same, but always different. Kumar isn't anti-AI. He says some generative AI is astonishing,
04:27some is a little scary, but a lot of AI companies are now hitting computational limits,
04:33pushing beyond what today's modern computers can handle. We're starting to see a lot of these
04:37companies choke on their own ambition. And that ripple effect hits the music industry too. Kato also
04:43isn't anti-AI. He actually likes using it as a collaborative tool, especially for new artists
04:50still building their sound. I think new artists and new songwriters especially, who are still kind of
04:57in that 10,000 hours phase, can use a platform like Suno, type in their lyrics, or feed it one of their
05:06beats that they made, and then just generate a ton of ideas that they can work off of. But legal
05:13questions get marky. Multiple lawsuits are being filed over who owns the rights to AI-created music.
05:19And they say that, okay, if you're on a pro subscription, you own the copyrights, right?
05:25But that's also like contingent on whether it can actually be enforced, because they know that
05:33they've trained their model off of the world's music. And on top of that, both Kato and Kumar
05:39agree that AI often lacks real human authenticity. The way that AI has been applied, and specifically the way
05:47that data has been stolen, the way that we are, you know, prompting, you know, prompting visuals
05:56without really much artistic practice involvement in that, I think is unsustainable and also is unlikely
06:03to in the long term really resonate with users, with consumers of content, of media.
06:08The Ipsos Deezer poll also found most listeners want clear labeling when a song is 100% AI.
06:15Half believe AI will play a major role in music creation within 10 years, but for the people
06:21who do this for a living, the concern isn't just technology, it's preserving the value of human craft.
06:27As a music producer of the last 15 years, I care deeply about this craft in this art form. And so
06:35when I see people who are able to kind of like play ball in our space now because of tools like AI
06:43and some of these other platforms, it makes me feel a certain type of way, you know, but at the same
06:49time, I'm never going to hate on someone that's trying to get their shine on.
06:53The future of music may be part human, part AI, or something in between soon. But as Harry Kumar put
07:01it, the future is coming, that's for sure. For more on this story and others,
07:05head over to san.com or download our mobile app. I'm Kennedy Felton with Straight Arrow News.
Be the first to comment
Add your comment

Recommended

12:47
Up next