Udio and the death of creativity

Founded in 2023, Udio is a generative artificial intelligence model that creates music from user prompts, with vocals and instrumentation. A free beta version of the platform was released to the public in April 2024. The program has received backing from a venture capital firm and endorsements from musicians such as will.i.am and Common. With this in mind, one may question how the music industry landscape may change with the introduction of artificial intelligence into creative processes.

In April 2024, more than 200 artists signed a letter urging tech firms not to develop AI tools to replace human creatives

Art, in any aspect, is ultimately an expression of the artists. Using artificial intelligence to replace this outlet defeats the purpose of artists experimenting, trying and failing in the natural artistic process of music creation. The current partnership between Universal Music Group (UMG) and Udio is highly concerning, as it encourages new artists to hasten their craft with the “aid” of generative AI while simultaneously discrediting artists who have been working their way up in the industry for years. Among artists currently signed under UMG are Taylor Swift, Billie Eilish, Ariana Grande, and many more, some of whom have taken a firm stance against the use of AI in music-making. In April 2024, more than 200 artists signed a letter urging tech firms not to develop AI tools to replace human creatives.

While some praise the platform’s ability to create realistic-sounding vocals, others raise concerns about potential copyright issues. Recently, Warner Music Group has settled their copyright infringement lawsuit with the platform, announcing the two companies will work together on a new AI music platform. Considering Udio operates as a large language model (LLM), using readily available information coded in the system’s database to generate this artificial content, it is hard to ignore the eventual copyright issues that will arise.

Udio’s generation of song lyrics relies on the LLM, while the generation of music itself has not been disclosed as of late. Considering that Udio agreed to launch a platform that would only feature “authorised and licensed music”, it still raises questions about how the model actually operates. With lyrics and even melodies sounding remotely similar to songs released by artists before, who would even stream AI music? It could become a platform for up-and-coming artists to experiment with their sound. Then again, with criticisms surrounding AI-generated music dubbing it as “soulless”, it is hard to imagine any artists normalising this route to develop their craft.

Fans of the artist took to social media to express their disappointment in her decision

As for the use of AI in music, Imogen Heap, for example, has developed an AI voice model called ai.Mogen. Heap released a new single back in October, utilising this AI voice. On her website, she writes about her concerns about the unethical training and environmental impacts of the technology, citing that the technology itself is not to blame. She urges her fans to do something with this technology rather than succumbing entirely to fear. In Imogen’s case, her consent was given, as she chose to train and use the AI model of her voice. She took a firm stance before it was done behind her back. However, fans of the artist took to social media to express their disappointment in her decision, citing the death of creativity by involving this technology in music creation.

When AI first entered the music industry, fans were quick to use it to generate covers of their favourite artists singing songs from other musicians, likely unaware of the repercussions at the time. When Lady Gaga released her latest album, MAYHEM, some fans pointed out that her song ‘How Bad Do You Want Me’ was reminiscent of Taylor Swift’s lyrical style and even her voice. It was then that fans quickly used AI models to generate Swift’s version of the song and even a music video of the two artists collaborating. While this situation was uncanny, other fans pointed out that using both the artist’s music and images was ill-advised for further training the LLM.

On the other hand, when The Life of a Showgirl came out, Taylor Swift had released several promotional videos entirely generated by AI. Although it was not used in Swift’s creative process in writing the album, it shows how normalised the use of AI is in the creative industry. Even if some fans were quick to jump to Swift’s defence, claiming that she likely had no idea that those promotional videos were being released, it still raises the question of how long it will take for this artificiality to leak into the creative process and how soon our favourite artists will succumb to using this technology.

Artists should jump at the chance to develop their craft on their own, with the help and feedback from peers, not an artificial model

Overall, while artists such as will.i.am, a firm believer in AI’s potential in music, continue to expand on this possibility, citing that artists can pull songs into existence with this tool, others argue otherwise. When all is said and done, it is essential to note that creating music (or any piece of art, for that matter) is never a linear process, but it is a rewarding one. Artists should jump at the chance to develop their craft on their own, with the help and feedback from peers, not an artificial model, and the use of AI essentially diminishes this notion.