Guest Perspectives

Through the Hall of Mirrors:
AI and the Expansion of Musical Imagination

Artificial intelligence has become one of the defining subjects in contemporary discussions of music. Much of the public conversation has revolved around familiar concerns: authorship, originality, copyright, and the fear that machines may one day replace human creativity. Brian Eno, writing in the Boston Review (December 2024), offered a thoughtful alternative to these anxieties. Rather than presenting AI as either a utopian breakthrough or an existential threat, he argued that it should be understood as another creative technology whose value depends entirely upon how artists choose to engage with it.  

I find myself in broad agreement with Eno's position. His essay rightly rejects both technological determinism and cultural panic. Like recording, synthesis, sampling, and generative systems before it, AI undoubtedly belongs within the continuing evolution of artistic practice.

Yet I also believe his historical framing, persuasive though it is, stops short of recognizing what is genuinely novel about AI. My own experience as a composer leads me toward a rather different conclusion. AI is valuable not because it composes, nor simply because it is another creative tool. Its significance lies in the way it transforms the composer's own relationship to musical imagination.  

Long before artificial intelligence emerged, composers had already developed countless methods for creating productive distance from their own ideas. Chance operations, algorithmic procedures, improvisation, tape manipulation, electronic processing, and computer-assisted composition all sought to interrupt habit and invite discovery. None of these techniques existed to replace the composer's imagination. They existed to enlarge it. AI naturally belongs within this lineage, but it also extends it in ways that distinguish it from its predecessors.

For me, AI is best understood as an instrument of exploration rather than production. It is not the composer; it is a catalyst that encourages the composer to hear differently. I often think of the process as looking into a crystal ball or wandering through a hall of mirrors. Musical material that I present to an AI system returns transformed-sometimes subtly, sometimes radically. Familiar gestures are reflected back with altered proportions, unexpected harmonies, unfamiliar timbres, or structural relationships that I might never have consciously imagined.

The result is never an answer.  

It is a question.  

What if the music went here instead?   

Eno has frequently argued that creative tools become most valuable when they introduce productive uncertainty rather than merely increasing efficiency. I find this observation compelling. Uncertainty has always been one of the great engines of artistic invention. Where I depart from Eno is in understanding the nature of that uncertainty.  

Eno largely treats uncertainty as a cultural condition-an openness that allows new creative practices to emerge. I see AI operating at a more fundamental level. It changes perception itself. Earlier technologies primarily amplified existing intentions or provided new methods for realizing ideas already imagined. AI often proposes relationships, continuities, and transformations that the composer had never consciously intended. It returns one's own musical material refracted through unfamiliar perspectives, creating the extraordinary experience of encountering one's own artistic language almost as though it had been composed by someone else.

This distinction is more than semantic.  

AI does not merely assist composition; it reorganizes listening.  

That, I believe, is what makes it historically distinctive.  

Every composer gradually develops recognizable habits-preferred intervals, harmonic tendencies, characteristic orchestrations, favored formal strategies. These habits become part of one's artistic voice, yet they can also become invisible constraints. AI possesses a remarkable ability to disturb these patterns without destroying their identity. It exposes possibilities already latent within the material while revealing paths that conscious intention alone might never have uncovered.  

Its greatest contribution, therefore, is not the music it generates but the dialogue it initiates.  

An unexpected harmonic progression may suggest an entirely new formal direction. An impractical orchestration may inspire an elegant solution. A peculiar rhythmic displacement may uncover structural relationships that had remained hidden. Frequently the AI's output itself is discarded, while the compositional path it reveals becomes indispensable. In this sense, AI functions less as a destination than as a compass.

Perhaps the greatest misconception surrounding AI in music is the belief that it seeks to replace the compositional act. My experience has been precisely the opposite. The composer's judgment becomes even more essential. Every suggestion must be accepted, rejected, refined, transformed, or completely reimagined. Taste, memory, intuition, craft, and aesthetic purpose remain irreducibly human responsibilities. AI cannot determine what is meaningful. It cannot recognize beauty, emotional resonance, or artistic necessity. It merely multiplies possibilities.

The composer alone decides which of those possibilities deserve to become music.

This also explains why I remain unconvinced by both the utopian claims of AI enthusiasts and the dystopian fears of its critics. Both sides remain preoccupied with the music AI might produce. They debate whether machines will become composers.  

To me, this is the wrong question.  

The more profound question is how AI changes the way composers hear.  

Its greatest contribution lies not in generating works of music but in expanding the field of musical imagination itself. It enlarges the possibilities available to both ear and mind before the hand ever commits an idea to notation or sound. Sometimes these discoveries are modest. Occasionally they are transformative. They remind us that composition is not simply the execution of preconceived ideas, but  an ongoing conversation with the unknown.  

Ultimately, AI enriches the compositional process because it returns the composer to composition with fresh ears. Its greatest gift is not automation but estrangement-the opportunity to perceive one's own musical thoughts from unexpected perspectives. Like peering into a crystal ball or walking through an endless hall of mirrors, AI reflects possibilities that were always latent within the music but had not yet come into focus.  

The real work begins after those reflections appear.  

It is then that the composer recognizes, rejects, shapes, or transforms them into something unmistakably personal. The composer remains the author. AI simply broadens the landscape through which the imagination may wander. It opens doors that might otherwise have remained unseen, inviting exploration without dictating destination.  

This is where my view ultimately diverges from Eno's. AI should not be understood simply as the latest entry in a long history of creative technologies. It represents something more profound. It is a technology of perception.

Its greatest innovation is not that it composes music, but that it enables composers to hear themselves differently.

In that sense, AI does not diminish creativity. It enlarges the space in which creativity can occur. Like every truly transformative development in the history of music, its lasting importance will not be measured by what it creates on our behalf, but by how profoundly it expands the horizons of human imagination.  

~ Robert Scott Thompson, July 21, 2026