Data Privacy

Cage, Schumann, Glass. Nando was obsessed with music and the history of music. When he went to study at a famous conservancy, he hoped to expand his horizons and build on the broad shoulders of many musicians he admired. So, he looked for ways to blend traditional composition with AI for his final junior year project. He had heard some of his classmates talk about AI tools, but he never looked into them himself. When he did his research, Nando found an AI tool that a musician could use to analyze their composition style and provide suggestions to create new melodies and harmonies that could accompany that composition. This he had to try!

The AI tool looks simple. He started by taking one of his recordings and uploading it to the platform. Within minutes, the platform offered Nando a range of samples that resonated with his musical taste. But it was long before he became concerned. A clause at the bottom of the page described their privacy and usage policy. He knew a little about AI and that it was built with data sets of music and samples from an extensive range of musical artists, but he then wondered whether or not the composition he just uploaded would also be used to create music for others.

Nando went back to the terms of service that he signed. Like many others, he didn't read what he signed up for, and he wondered, "What exactly did I sign up for? Could my compositions be used to help somebody else generate music or samples? It reminded him of a podcast he often listened to that featured an episode on digital ethics and mentioned how few of us know how our data will be used. He also remembered a story he heard from one of his classmates about how a song uploaded to one of these websites got leaked to the public. That could be his work! He became dubious of uploading anything else to the website.

As Nando continued to research the issue, he found musicians who actually embraced the idea of their music being used in AI training. They told him it was part of music’s evolution, similar to how jazz musicians would riff off each other during a set. They told him it’s a way of collaborating, pushing the boundaries of musical creation, and that it would probably lead to new genres of music. Some musicians even released a Creative Commons license that grants AI developers the freedom to use their compositions to train new AI tools. They believe this could democratize music creation and lead to more diverse and innovative compositions.

On the other hand, if everyone's music and data became part of a training set, would it lead to homogenized music? And would artists’ unique voices be lost in a blended sea of AI-generated music? He was also concerned that others might get access to his private data through his metadata. These questions added another layer of complexity to his already conflicted feelings about using the AI tool.

He talked with his friend about all this, someone who knows a lot about these AI systems. They explained that some of these music data sets only have musical samples from a small demographic of artists. His friend suggested that this makes it more likely that only a subset of musical history will go into generating AI-generated creations. That same friend pointed out that copyright is another issue he should consider. Nando wondered if he would violate copyright laws if the AI tool's music was based on protected music illegally uploaded to the data set. As someone who considers himself to be very private, he was concerned that this might expose his true identity. All these factors made him uneasy about continuing to use the tool, at least until he has more clarity on these issues.

What do you think?