Towards the Generation of Musical Explanations with GPT-3

TL;DR

This paper explores GPT-3's ability to generate textual explanations of musical decisions, highlighting current limitations due to lack of specialized training data, and emphasizing the importance of such explanations for human-AI musical collaboration.

Contribution

First study assessing GPT-3's capacity to produce musical explanations, identifying data scarcity as a key barrier to understanding musical decisions.

Findings

GPT-3 struggles to accurately explain musical decisions.

Lack of explanatory training data hampers AI understanding of music.

Enhancing datasets could improve AI's musical interpretability.

Abstract

Open AI's language model, GPT-3, has shown great potential for many NLP tasks, with applications in many different domains. In this work we carry out a first study on GPT-3's capability to communicate musical decisions through textual explanations when prompted with a textual representation of a piece of music. Enabling a dialogue in human-AI music partnerships is an important step towards more engaging and creative human-AI interactions. Our results show that GPT-3 lacks the necessary intelligence to really understand musical decisions. A major barrier to reach a better performance is the lack of data that includes explanations of the creative process carried out by artists for musical pieces. We believe such a resource would aid the understanding and collaboration with AI music systems.