Music composition in the 21st Century: exploring concertgoers’ aesthetic response to AI-generated music.

Abstract

We live in the information age where digitisation and computational technology have become integral and indispensable to our daily activities. Artificial intelligence (AI), quantum computing, and other such technologies increasingly impact and disrupt our lives as we connect with our world. Within the arts, a field once dominated by human creation, we now experience a penetration of AI and deep learning technologies. The researcher, a practising musician, became interested in how our ubiquitous interaction with AI technology affects our decision-making and how it relates specifically to the field of music composition. The manifestation of AI’s impact on music-making was met with the researcher’s excitement and trepidation. Given the researcher’s apprehension, he proposed investigating (1) the quality of AI creativity in the field of music composition and (2) how transparency of this AI creative employment affects aesthetic judgement. He designed his research using a mixed methods approach, comprising a quantitative phase in the form of an online questionnaire (based on the original AESTHEMOS instrument), followed by a qualitative phase of in-depth interviews. The researcher’s objectives were twofold: (1) to establish if a sample of concertgoers could discern aesthetically between compositions generated by humans and AI and (2) how knowledge of AI use during the compositional process affects our aesthetic appreciation of the artefacts. The researcher partly hypothesised that participants could not discern aesthetically between human and AI-generated compositions because of current available AI technology (through machine and deep learning). However, when AI employment is disclosed, aesthetic responses to compositions yield a negative response. To test his hypothesis, the researcher engaged thirty concertgoers in Amsterdam, The Netherlands, to evaluate aesthetically five symphonic works via an online questionnaire. During the follow-up interview process, the AI generation of two of the five compositions was disclosed, and general attitudes toward AI creativity was probed. Using data analytic tools such as the Mann-Whitney U test, the researcher confirmed his hypothesis and concluded that participants interact aesthetically with AI-generated compositions if they appear to be human-composed. Transparency of AI involvement, however, affects the aesthetic value of AIgenerated compositions. As AI weaves itself deeper into the human story, the familiarity of AI creativity will profoundly affect our notion of creativity, meaning and art creation of the future.