Bop or bot? New study reveals the artists who are losing out on AI-generated music the most
The availability of artificial intelligence and the widespread prevalence of AI-generated songs and covers has not only brought significant changes to the music industry but has also sparked controversy. As AI continues to blur the lines between genuine and synthetic creativity, fans are finding it increasingly challenging to distinguish between songs crafted by human musicians and those produced by machine learning models. With an estimated 1,630,000 AI covers currently on YouTube alone, this technological takeover has begun to raise concerns about the authenticity of music, whether human creativity is important, and the financial implications on artists.
Concerningly, studies have found that 2 in 3 Americans cannot recognise the use of AI in creating or completing music compositions [1], underscoring the increasing sophistication of AI’s ability to replicate human artistry. As such, many high-profile musicians, such as Billie Eilish, Nicki Minaj and the estates of Frank Sinatra and Bob Marley, have begun to view AI as a threat to jobs and artistic integrity and an “assault on human creativity” [2]. Many fear substantial revenue losses as AI-generated tracks flood the market, often mimicking popular styles and voices.
To understand the scale of threat and revenue loss, musicMagpie conducted a comprehensive analysis of 1,500 AI-generated covers and songs to understand the amount of streams, number of videos, and the popularity of particular AI artists. The findings revealed that some of the industry's biggest names are significantly affected, with AI-generated tracks amassing millions of streams and effectively syphoning off potential earnings from the original creators.
Ranking the most streamed AI-generated artists
Current icons
The study revealed that the most streamed AI-cover artists are the K-Pop girl group Blackpink, followed by Justin Bieber and Kanye West. In the playlist analysed, AI-generated content from these three artists alone accumulated over 33 million views.
Overall, AI-generated K-pop artists are the most streamed, with Korean bands and solo artists making up 35% of the top twenty artists. Blackpink led by a considerable margin with over 17.3 million views. Their most popular AI-cover songs were BabyMonster originals ‘Batter Up’ (over 2.5 million views) and ‘SHEESH’ (1.88 million views). Further K-pop AI royalty came in the likes of Jimin (1,683,898 views), Jennie (950,577 views), Stray Kids (677,270 views), Twice (659,699 views), NCT Dream (447,605 views) and Enhypen (428,654 views).
Western artists were also a target for AI-generated content. With over 13 million streams, Justin Bieber is the second most affected artist–his widespread popularity and extensive discography likely make him a frequent target for AI replication. His top streamed AI covers included ‘Nothing’s Gonna Change My Love For You’ by George Benson (10.1 million views) and Glimpse of Us by Joji (118,428 views).
AI-generated Kanye West holds third place overall with approximately 3.4 million views, with the “Kanye Ft. Playboi Carti” cover of ‘Somebody That I Used to Know’ raking in a whopping 2.6 million streams. Other top-ten contenders included Playboi Carti (2,675,926 views), Lana Del Rey (2,527,881 views), and Linkin Park (2,294,047 views).
Legacy voices
AI technology has extended its reach beyond contemporary music, tapping into the voices of legendary artists who have passed away. These AI-generated renditions provide fans with a unique, albeit controversial, way to experience the iconic musicians from the past.
Topping the leaderboard for the most-streamed “legacy voice” was Frank Sinatra, who had a staggering 8,923,342 views, defining him as a major figure in the realm of AI-generated music. Following this was the legendary Queen frontman, Freddie Mercury, who has amassed a whopping 3,555,009 views through AI-generated content. In third place was Kurt Cobain, whose grungy vocals have been reworked to accumulate 1,878,043 streams. The King of Pop, Michael Jackson, was ranked fourth (1,820,862), followed by Hank Williams (1,750,470), Johnny Cash (1,210,782), Kishore Kumar (556,033), Prince (482,214), Elvis (161,354) and The Beatles (63,029).
Imaginary Hits
AI technology has also ventured into the realm of fictional characters, celebrities, and cartoons, producing a fascinating, albeit strange, array of music.
Our analysis found that SpongeBob SquarePants is a standout in AI-generated music, with the character amassing a staggering 10,247,793 views (should we call for a collaboration with Bieber?). SpongeBob’s most in-demand solo track was a cover of Don Maclean’s ‘American Pie’. However, his collaborations with castmates, for example, the cover of ‘Someone You Loved’, which includes SpongeBob, Patrick, Mr Krabs, Plankton and Squidward, was more successful (2,996,738). Plankton followed shortly after SpongeBob with 5.1 million views, his top hit being a cover of Maneskin’s ‘Beggin’’.
Donald Trump was another AI-loved cover artist, reaching 4,039,923 views, surpassing his opponent, Joe Biden, who had 3,285,404 streams. AI-Trump’s greatest success came from a cover of ‘You’re Welcome’ from Moana, which surpassed half a million views. AI-Biden’s top hit was a cover of ‘Boy’s a Liar’ by PinkPantheress, which saw 3,274,927 streams. The two also tag-teamed and achieved 2,908,760 views of a cover of Jay-Z and Kanye West’s ‘Ni**as In Paris’
Ranking the most streamed AI-generated covers
When it comes to the most streamed AI-generated covers, Billie Eilish’s ‘bad guy’ tops the list with a mind blowing 20.5 million views, covered by an AI-generated SpongeBob, an electric toothbrush and Helen Keller. Following closely is Elvis Presley’s iconic ‘Can’t Help Falling in Love’ amassing over 18.5 million streams, voiced by the likes of a machine-learnt Lana Del Rey, Freddie Mercury and Kanye West. Rounding out the top three is ‘Redbone’ by Childish Gambino, reimagined with AI-generated voices like Michael Jackson and Freddie Mercury.
The original artists most plagiarised by AI
Certain artists saw multiple covers of their songs, and topping this list was Kendrick Lamar whose diss track ‘Not Like Us’ topped the highest number of AI-generated covers, covered by the likes of “Donald Trump”, “Homer Simpson” and “Drake”.
Close behind is British popstar Harry Styles. His track ‘As It Was’ has been a popular choices for AI covers, with renditions performed by the AI-generated voices of Olivia Rodrigo, The Strokes and Dua Lipa.
Frank Sinatra secures the third spot with to his anthem ‘My Way.’ This song has seen a significant number of AI covers, including the voices of Homer Simpson, Arthur Morgan and Joji.
Revenue loss
The surge of AI-generated music is posing unprecedented challenges for artists, particularly due to the lack of licensing from copyright holders. Many artists are seeing significant drops in their earnings, which would have otherwise been generated from their original work.
Our analysis estimates that there are approximately 1.63 million AI-generated covers on YouTube, with a median average of 2,082 views each. By converting these views to Spotify plays, musicMagpie has inferred that the total revenue loss for artists currently stands at a staggering £10,662,897.55 ($13,571,380.00).
Topping the list of affected artists is Blackpink, with an estimated revenue loss of £376,184.85 from 1,463 analysed videos. Michael Jackson follows closely, with a loss exceeding a quarter of a million pounds (£262,265.09). In third place is BabyMonster, who have missed out on an estimated £246,499.79 due to AI-generated covers.
Can you tell AI from human?
To better understand the threat posed by AI in the music industry, we conducted a study involving individuals aged 21-60, aiming to gauge their ability to distinguish between human-created music and AI-generated content. Firstly, we asked participants “do you think you could recognise AI in a song?”, then they were presented with two tracks: an AI-generated Drake song and a lesser-known, authentic Drake track.
Despite 72% of participants expressing confidence in their ability to recognise AI-generated music, the reality told a different story. Nearly 1 in 2 participants (49%) were unable to correctly identify the AI cover, demonstrating how convincingly these technologies can replicate human artistry.
Breaking down the data further, the study revealed Gen Z were the most incapable of recognising AI with 52% of participants failing to grasp to the real Drake song, despite growing up in a digital age. The older generation (Gen X) were split evenly, with 50% of Gen Xs being fooled by AI. Millennials had the best grasp with only 45% misled by AI.
These findings highlight a growing challenge in the music industry: as AI technology becomes more sophisticated, music-lovers across multiple generations are struggling to discern between what is real and what is artificially created. This blurring of lines not only raises questions about the future of music’s creation but also poses significant implications for artists, record labels, and the industry as a whole. If nearly half of listeners can’t tell the difference between a human artist and an AI, what does this mean for the value of human creativity? How will this affect the way we create, perceive and appreciate music in the years to come? These are questions that the industry must grapple with as AI continues to evolve.
Methodology:
To get a snapshot of the most plagiarised and streamed AI artists, over 1,500 AI covers on Youtube were scraped and cleaned to provide an overview of the AI generated music industry. This data was then analysed to find the most viewed song covers, the most used original songs and the most AI generated voices within the music industry. To calculate a snapshot of the potential revenue loss, the number of views was used in conjunction with the cost per stream on Spotify ($0.004 per stream).
To work out how many people could recognise an AI generated artist, musicMagpie conducted their own study among 47 individuals, aged 16-60, using two songs. Drake's original song “No Telling” vs an AI cover “Meet the Grahams”. After playing both songs we asked each participant which they believed to be the real artist and which was the fake AI generated artists and analysed the results.