Reading Response

02/04/2023

Vic

In honor of the AI and AI ethical focus, I decided to use chatGPT to help me write this article to supplement my own perspective and thoughts. All chatGPT-produced responses will be quoted and labeled as clearly as possible. 

As a quick introduction to the topic, “The article introduces MusicLM, a state-of-the-art model for generating high-quality music from text descriptions. Music generation from text inputs has been an active area of research in the field of artificial intelligence and machine learning. The authors aim to provide a solution to this problem by presenting MusicLM, which generates music that remains consistent over several minutes and outperforms previous systems in terms of audio quality and adherence to text descriptions.” (chatGPT 2023)

MusicLM is a research project by google exploring the use of text descriptions to generate music. While this is an exciting advancement to some, it is a troubling omen to others. Several points of concern can arise from this topic. For example, how could this affect the jobs of current musicians? What kinds of rights would be put in place on this type of generated music if the sources come from others? 

I personally feel that with the way the music industry currently is, and the strictness it has on music use currently, it is unlikely that much if any money could be made off of this product. Whether we agree with it or not, the companies and artists who own any source materials will be claiming their own songs or preventing them from being used, and any used would likely be highly monetized. I assume this is based on the current state of music and its use on youtube. For youtube ads are placed on copyrighted material, monetization of the video creator isn’t allowed, and permission has to be granted in order for the music to be uploaded and included in a video. Any source material would have to be from either artist who authorizes the use, or from in-house musicians who make music specifically for the model to learn off of.  

Even if the user is inputting sample material that happens to come from a copyrighted source, it wouldn’t surprise me to see the song recognition technologies google developed for humming songs being used to recognize and prevent unauthorized copyright infringement on their platform. 

Besides the issue of monetization and unauthorized use of commercial music, I do think that it has great potential for creativity and idea-making. That is of course if there is a wide enough range of sounds for it to work off of that it can create diverse ideas. But having your own existing idea and trying to test different instruments, styles, rhythms or arrangements does have its own appeal as a quick and useful tool for music production. While immediate use for songs created by the software would most likely be monetized by the company itself, using it on a small scale to listen to different ideas seems like a non-invasive method of exploring the possibilities. 

One of the more troubling possible issues with this technology is the potential replacement of musicians and jobs in the music industry. While this technology is exciting and has many potential applications, I think its audience for use will not be the same audience that wants musicians specifically to begin with. Much like stock photos provide images to companies, I think that this kind of auto-generated music will have used for large-scale groups and companies that want easy-to-find materials, for generic projects. However, any high-priority or unique projects will still want personally made elements with an unmistakably human touch.