The Artificial Lyricist: Prototyping an Interactive Opera for Humans and Machines

To probe into the artistic possibilities of Artificial Intelligence (AI) generating operatic elements, we have engaged in a practice-led inquiry as designers and artists. We took the process of prototyping a chamber opera with interactive features as our starting point, exploring how AI could work as a source of stimuli in opera productions, and how prototyping might add to the design and evaluation of AI integrated into an operatic context. Our core idea was to use AI to generate personalized lyrics in a dynamic opera libretto, based on input to the AI from individual visitors attending the operatic event. In response to visitors, the opera singer read and rendered the AI-generated text artistically in real-time and in a karaoke-like manner in accordance with the musical framework and the embodiment of the operatic character. Hence, the concept comprised both human-to-machine, machine-to-human, and human-to-human interaction. While some design elements stimulated the participating opera visitors, others stimulated the performing opera singer, supported by an informational framework composed verbally through a prototypical manuscript, audially through a prototypical accompaniment, and visually through a prototypical setting. The prototype was play-tested and evaluated in relation to our artistic intentions. The AI-based prototype added parameters to the artistic process such as sketching narrow AI for the operatic format and evaluating articulated qualities. Furthermore, we suggest that the conceptual artwork can be seen as an example of a contemporary turn in operatic evolution, with human performers forging not only bodily, but also intellectual relationships with machines.