Mark Zuckerberg Says AI Music May Seem “Soulless” Despite Technical Precision

Mark Zuckerberg might be fluent in code, but even he admits there’s one equation AI still struggles to solve: the soul of a song.  

Appearing on the Track Star show, the Meta founder delved into the ongoing debate surrounding artificial intelligence's role in the creative arts. As he identified songs by some of his personal favorite artists, including Daft Punk, during a music quiz, the conversation naturally drifted to whether or not AI could ever truly rival human artistry. 

"AI will probably be able to produce technically interesting music," Zuckerberg said, "but it may sometimes feel a little soulless because it lacks the other parts of the human connection."

His comments come at a time when the scourge of AI-generated music continues to spark passionate discourse across the music industry. Major labels have ramped up their legal defenses, like the explosive lawsuits against generative AI startups like Suno and Udio, which were accused of using copyrighted songs to train AI models without permission.  

For Zuckerberg, who has advocated for open-source AI technology, the creative realm poses a conundrum. In recent months, he joined Spotify CEO Daniel Ek in backing open AI models as a means to democratize innovation and avoid centralizing power among a few tech giants. AI music, however, poses a unique challenge. With the lack of human experience baked into AI algorithms, songs created by machines might technically impress but lack emotional resonance.  

Zuckerberg's recent personal brush with music even includes recording a quirky cover of "Get Low" for his wife Priscilla Chan, but when it comes to AI music's future, it seems he himself is on the fence. With ongoing regulatory concerns and increasing pressure from artists to protect their work from AI encroachment, the music world continues to grapple with whether AI can ever truly find harmony with human creativity. 

View the original article to see embedded media.

You may also like...