The Beethoven Decima and Artificial Intelligence: What is Creativity?
On October 9, the Beethoven Orchestra of Bonn performed, for the first time, Beethoven’s Tenth Symphony in its entirety. A surprising event for classical music experts: the Tenth Symphony had remained unfinished after the German composer’s death in 1827. So who completed this composition? A totally unexpected author: a computer!
You can listen to the composition here.
This project was handled by a research team from the Art & AI Lab at Rutgers University in New Jersey.
Ahmed Elgammal, who leads this ambitious research project on computational creativity, has been working on it since the early months of 2019. This project, which has faced a series of challenges, set itself the exciting task of replicating a style more complex and daring than previous research that had focused on harmonizing melodies in the more regular and “mathematical” style of Bach.
All this raises a question: can an Artificial Intelligence program be compared to what we commonly define as “creativity”?
Computational Creativity and Human Creativity
Computational creativity is defined as “the computational analysis and/or synthesis of works of art, in a partially or completely automated manner” [1].
The generative models used for Computational Creativity, such as that of Elgammal’s team, “create” an output based on a training dataset through algorithmic processing. In the literature, a generative model can be evaluated based on two parameters:
“1) whether it can generate examples that appear to be drawn from the same distribution as the training dataset, a concept known as fidelity, and
2) whether the examples are sufficiently different from those seen during training, in other words, diversity.” [2]
The flow of the creative process in a computational creativity system is therefore determined purely syntactically.
This is inherently different from what happens with human creativity, a process in which the semantic aspect is crucial in processing information—a process not strictly of a logical-mathematical nature but partly based on intuition and emotion.
In conclusion, the development of Artificial Intelligence in the field of Computational Creativity shows us that machine learning “is not alien to the world of the arts” [3]: however, creativity—as a concept subject to interpretation and suggestion—remains in the hands of human composers.
Federico Mandelli, Federico Lagiannella
To learn more: https://www.ilsole24ore.com/art/l-incompiuta-decima-sinfonia-beethoven-portata-termine-grazie-all-intelligenza-artificiale-AEYmrXm
Notes:
- J. D. Fernandez, F. Vico, (2013) “AI Methods in Algorithmic Composition: A Comprehensive Survey”
- Naeem M. F., Oh S. J., Uh Y., Choi Y., and Yoo J. (2020), “Reliable Fidelity and Diversity Metrics for Generative Models”
- R. F. Cadiz, A. Macaya, M. Cartagena, D. Parra (2021) “Creativity in Generative Musical Networks: Evidence From Two Case Studies”