Playlist: Discussing AI-generated classical music with ChatGPT

My conversation with ChatGPT 3.5 about AI-generated music does not start too convincingly. To break the ice, I inquire about a recommended piece of classical music crafted by artificial intelligence. ChatGPT promptly suggests Symphony for a Better World by Aiva.  

“Aiva is an AI music composer that has produced various classical pieces, and Symphony for a Better World is a notable composition”, explains ChatGPT with confidence. “Give it a listen and see how AI can create captivating classical music!”  

The hitch? Such a composition does not actually exist. Aiva does exist, though, and is among the most extensively used AI platforms for music creation. Therefore, let’s delve into another orchestral piece composed via Aiva, wittingly titled Symphonic Fantasy in A minor: I am AI.

The composition incorporates a fascinating interplay: AI-generated music being brought to life by a live orchestra. What are ChatGPT’s thoughts on this dynamic?

“It’s a beautiful synergy that showcases the potential for technology to inspire and collaborate with human creativity, offering new avenues for musical expression and exploration. The result often intrigues audiences by presenting a harmonious blend of the technological and the human elements in music.”  

ChatGPT’s highly optimistic perspective contrasts sharply with those who perceive AI as a threat to musicians, at least on recorded music. How do you personally perceive a composition when aware that human musicians are performing music generated by a computer? How does it differ from listening to a performance of music by a human composer?  

“Beethoven’s Tenth” 

Beethoven left behind several sketches for a tenth symphony. Efforts to reconstruct the work have involved scholars, including recent attempts aided by artificial intelligence. Let’s compare two examples. Musicologist Barry Cooper compiled a hypothetical first movement in the 1980s, while the Beethoven X project recently relied on AI for the editing process of the third and fourth movements, followed by a performance and recording by the Beethoven Orchestra Bonn.

To be fair, these two reconstructions are not entirely comparable since they involve different source material and reconstruct different movements. However, they do offer intriguing insights into attempts to reconstruct music from sketches.  

Could AI potentially aid this process in the future? ChatGPT holds an optimistic view of such endeavors, highlighting AI’s potential to significantly aid in completing unfinished works and foster collaboration and inspiration between AI and humans. Ideally, human composers, programmers, and scholars collaborate with AI systems to “leverage technology to inspire new musical creations or interpretations.”  

While ChatGPT predominantly champions AI’s creative potential, it prompts me to reflect on AI’s limitations in expressing emotions and offering subjective interpretations.  

“Some argue that while AI can imitate styles, it might lack the true emotional depth and personal touch of the original composer”, ChatGPT elaborates. 

These instances once again present a captivating interplay, challenging the boundaries of creativity and expression. What defines “originality” in these reconstructions? Moreover, what is the creative influence of the AI’s creator?

Non-transparent referencing?  

Beyond reconstructions, AI-generated music is perhaps most recognized for its imitations of existing compositions. Let’s explore another example, this time not stemming from sketches but offering an entirely new composition: a Mazurka ‘after Chopin’. It is worth noting that this is a relatively early instance, dating back to 1997 – a time when artificial intelligence was scarcely considered a creative force.  

In this Mazurka, the model composer is explicitly identified. In the majority of AI-generated music, however, this information is not disclosed - despite the fact that the final composition often reflects a distinct model that serves as a reference for the AI-generated piece.   

Consider this orchestral composition, Symphonic Fantasy in A Minor: Among the Stars generated once again by Aiva. Many have noted haunting similarities with John Williams’s film scores. When I raise this issue with ChatGPT, it acknowledges the importance of being “transparent about the origin of the music”. For ChatGPT, this concerns matters of copyright and legislation.  

“Addressing copyright issues related to AI-generated music often requires a nuanced understanding of copyright law, including considerations for fair use, derivative works, licensing agreements, and the evolving nature of technology's role in creative processes. As AI continues to evolve in music composition, legal frameworks might need to adapt to encompass these new forms of creativity while protecting the rights of living composers and their intellectual property.” 

Simultaneously, we grapple with a profoundly human aspect. It’s simple to point fingers at AI for ‘plagiarism’ (or, more accurately, at the human programmers who trained the AI using copyrighted material). Yet, conscious and unconscious acts of emulation, pastiches, homages, inspiration, influence, and even copyright infringements have long been facets of the human creative sphere. How do these actions differ from one another, and how should they be assessed in the case of AI-generated music? 

Technology as a creative force 

Let us finally pivot to an entirely different angle on music and technology by experiencing Isao Tomita’s synthesized rendition of Debussy’s Clair de lune.  

In 1974, when Tomita released his Grammy-winning collection, where he reinterpreted Debussy using a synthesizer, his approach was lauded for its pioneering use of technology. How does the utilization of AI differ from such an approach? And does AI hold the potential for garnering similar acclaim in the realm of human creativity?  

“Absolutely!” says ChatGPT. “Isao Tomita used technology as a means of artistic expression, transforming classical compositions into something innovative and unique. Similarly, music created by AI deserves admiration for its innovation and the exploration of new artistic frontiers.”  

ChatGPT also believes that admiration for AI-generated music should arise from acknowledging the ingenuity of leveraging technology not merely as a tool but as a collaborator in the creative process.  

“Ultimately, the integration of AI in music can lead to a more nuanced appreciation of human creativity, emphasizing its distinct qualities while also highlighting the capabilities and limitations of technology in the realm of artistic expression. It’s an evolving conversation that continually reshapes our understanding of creativity and the human experience.” 

Consider if this reinterpretation was not crafted by Tomita but by AI. How might your evaluation of the creative elements within the piece change? Or would it change at all? 

Language revision: ChatGPT