We agree that artists should get paid if another artist uses their work, so why should it be any different when it comes to AI?

AI is reshaping the music industry, and independent artists are beginning to wonder how their intellectual property is being treated. A whitepaper by Professor Daniel Gervais, an intellectual property law expert, seeks to address a key question: should artists be compensated when their work is used to train AI systems? This idea, if implemented, could bring about a significant change in how musicians, particularly independent ones, are rewarded in the age of burgeoning AI.

It can sound nefarious, an artist’s work being reused by AI without care or consideration. But, the thing is, AI is getting used by people in this way, and many of those people are artists themselves. In fact, a survey this year found that over half of young creatives use AI in their creative process. Other surveys suggested that around 25% of producers use artificial intelligence in music-making. Although it is partially artists using AI, which is then trained on other artists’ work, there is still growing concern around what this means for the artists whose work the AI is trained on.

What is Gervais proposing?

Gervais argues for the introduction of a new legal framework that grants artists financial compensation when their work is used in AI training. To understand why this is being proposed, it’s key to have an understanding of how AI models generally work.

Currently, AI models rely on massive datasets, often made up of copyrighted material, to create new content. The issue is that creators don’t receive payment when their work is used in this way, despite its role in shaping the output of these AI tools. Gervais suggests, “It is essential to develop a system that recognizes and rewards the creative input of artists when AI is trained using their work.”

Under this proposal, artists would receive compensation when their work contributes to the training of generative AI models. This goes beyond the initial licensing fees or copyright royalties that they might receive for distributing their work in traditional formats. Instead, it introduces the idea of ongoing repayment whenever AI-generated content is created based on the original work of musicians. After all, if an artist uses another artist’s work, the original artist (should) get remunerated , so why not the same with AI?

How would paying artists work?

The idea certainly is appealing to artists, but putting it into practice would need big changes in regulations. Gervais points out that tracking how AI models are trained and figuring out which music or sounds were used is complicated. “We need a system that can accurately monitor the use of copyrighted material in AI training,” he says, highlighting the need for transparency.

This would likely involve setting up systems to log and report which artists’ works are used and how often. AI developers would then need to pay artists accordingly, creating a new way for musicians to earn money.

Challenges to overcome

There are, of course, challenges to making this proposal work. One big issue is tracking how an artist’s work is used in AI training datasets. AI systems use vast amounts of data from many sources, making it hard to determine which content influenced the AI.

Another issue is figuring out fair compensation. How much should artists be paid when their work helps train AI? And who decides these rates? Gervais acknowledges in his proposal that these questions need to be answered to create a fair and workable system.

What does this mean for the future?

Gervais’ proposal is an important step toward fairer treatment of musicians in the AI age. If it happens, it could lead to significant changes in how artists are paid, ensuring they benefit from their contributions to AI-generated content.

For independent musicians, this could mean a valuable financial boost, helping them continue creating in a changing industry. As Gervais says, “The future of music must include protections for creators, even as AI becomes an integral part of the creative process.”

But, a proposal does not necessarily mean this will happen. It could be a long time before we get any clarity on how AI-generated music and content is treated in a legal sense. With so many new and ongoing lawsuits around the topic, the outcome and answer to this overall question is currently anyone’s guess.