Internet Fooled by AI-Generated Kendrick Lamar Diss Track
Good Kid, Mad AI
A diss track featuring the apparent vocals of rapper Kendrick Lamar made its rounds on social media earlier this week, escalating the beef between him and Aubrey "Drake" Graham.
Now a 23-year-old musician who goes by the moniker Sly the Rapper has come forward, alleging he's behind the viral track, which was titled simply "Freestyle." And guess what? He says it was AI-generated.
That's impressive, because it fooled plenty of people into believing it was the real thing.
"I thought people were going to know that it was AI," Sly told Complex. "Although it did sound real, it sounded like an old version of Kendrick Lamar."
"The style sounded kind of dated, so I just thought that would be a giveaway, but apparently it wasn't," he added.
To Pimp a ButterflAI
The use of AI in the music industry has sparked a heated debate over copyright, music piracy, and musicians' rights.
The news comes after a song that used AI-generated vocals to imitate Drake and Abel Makkonen "the Weeknd" Tesfaye went viral last year. Incensed by the song's success, record label Universal Music Group forced Spotify, Apple Music, and other music streamers to take the song down.
The latest AI-generated diss track adds plenty of confusion to an already messy beef between Lamar and Drake. On Friday, Drake officially released the studio version of "Push Us," in which he took aim at Lamar.
The latest diss track also highlights just how easy it has become to generate believable-sounding vocals of famous musicians.
Sly told Complex that he used an AI software called Jammable — formerly known as Voicify AI — for his latest track, which allows users to create AI covers "in seconds." A quick search on the company's website reveals several user-submitted AI models that can be used to imitate Lamar's vocals.
Even YouTube has teased a new feature called "Dream Features" that's designed to allow users to create AI-generated songs using the licensed voices of famous artists including Demi Lovato, John Legend, and Troye Sivan.
Despite initial skepticism over Sly's claims, the rapper provided Complex with proof in the form of timestamped videos and audio, including isolated audio tracks.
The musician, though, has some mixed feelings when it comes to the tech.
"A lot of people are scared of AI, rightfully so," he told Complex. " I understand where the fear comes from, but AI isn't as advanced as people think that it is."
"There’s a lot that AI can't do," Sly added. "It still can't mimic cadences. It can't mimic emotions."
More on AI in music: YouTube Launching AI Tool That Clones Singers' Voices