Can Artificial Intelligence Create Original and Convincing Art?
ISEF · 2020 Systems Software
Overview
My goal is to determine if artificial intelligence can create art that is both original and comparable to human-created analogues. To determine this, I decided to use music as the artform, as it does not require external hardware as might be required by painting or sculpture, and it is a complex process. I will use the deep learning library Keras in conjunction with the open source TensorFlow software to build my A.I. As input, I will use the most notable classical music compositions in history, in midi form. My expectation is that my A.I. will discern the rules for creating these compositions and use the rules to create its own compositions. The A.I. will produce output in text data form, which I will convert to midi form. I will then apply an orchestral VST to the midi file in a separate audio program to emulate the sound of the appropriate symphonic instruments. I expect that the results will be original compositions that are free from errors (e.g. out of key notes, unintentionally out of rhythm sections, etc.). Broadly, my interest is in using A.I. to solve complex problems that require creative solutions, and not music per se. I hope to use my results to investigate other areas where an A.I. might benefit humanity through original ideas (e.g. risk analysis and mitigation, novel engineering designs, and countermeasures for the spread of disease, for example).
Competition history
- ISEF 2020
Resources
Related projects
ISEF · 2024
Creating an Artificial Intelligence Neural Network to Critique Expressive Piano Performance
ISEF · 2022
Analyzing the Musical "Genome" With Neural Networks
ISEF · 2021
Can We Differentiate Between Computer Generated and Human Art? Spot the Fake!
ISEF · 2025
Identification of Musical Key, Mode, and Style Using Various Supervised Machine Learning Methods
Closest projects by meaning, across every fair and year in the corpus.
Source: Regeneron International Science and Engineering Fair