The misuse of generative AI in music
Large Language Models (LLMs) have experienced a massive boom in the last couple of years. Their use has spread across all industries, creating a surge of multimedia content generated through Artificial Intelligence (genAI). After the initial wave of excitement, more and more institutions and organisations have come forward to denounce the negative impact of this technology in multiple aspects.
One of them has to do with creative ownership. In order to function, LLMs must be fed gigantic amounts of data. So far, there has not been adequate regulation about the origin of said data, nor how the human creators of it should be recognised and compensated. In the case of music, this has resulted in the work of hundreds of thousands of composers and recording artists being fed to LLMs without consent. To aggravate things, the musical output generated by LLMs is fast to produce and cheap to sell, which has degraded or outright eliminated jobs for musicians.
Updating the existing legislation on copyright is essential to face this crisis. In the UK, the Musicians Union has been campaigning to achieve that:
View this post on Instagram
To learn more, you can read the full briefing by the MU on their website.
In regard to what you can do to help with this issue, bear in mind that LLMs cannot function without the data that, so far, has been fed to them without consent and violating intellectual property. This means that, by using software like ChatGPT, Gemini, Midjourney, DALL-E, Copilot, Grok, AIVA, etc. you are supporting systematic plagiarism and hurting other creatives. Don't use genAI to create images for your promotional material or album covers, don't prompt any LLM to write your e-mails or social media posts. Just as you wouldn't want a song generated by one of these tools with your music as base data to be licensed while you struggle to find work, the illustrator whose art was stolen to train an LLM, so it could generate an image for your Instagram post, would rather have you contact them directly for a true artistic collaboration where talent and dedication are properly recognised.
In addition, it's important that you're aware of the fact that LLMs are not reliable sources of information. They don't have the capacity to distinguish right from wrong, and are prone to hallucinate facts and events that don't exist or haven't happened. If you want to learn something, or catch up with the news, go to human sources instead of "asking ChatGPT". This will get you veritable information of good quality, prevent you from hurting yourself and others by following false advice, and keep you connected to other humans in a network able to support your life and career.
Until genAI is properly regulated, and the humans behind the data to train it are rightfully remunerated and credited, every use of it is a form of misuse!