Grammy Chief Harvey Mason Clarifies New AI Rule


When Recording Academy announces latest update to Grammy Awards rules Last month, there was a rule that had to be simple, but it was actually nothing more than this: It took about 202 words to explain it. the emergence of artificial intelligence, only music truly created by humans will be eligible to be nominated for or won a Grammy.

Yes, recordings using one of multiple types of AI technology will be fine, but mainly words and melodies written by artificial intelligence — also known as generative AI — it is not. “We will not nominate or award an AI computer or someone who drives AI.” Registration Academy CEO Harvey Mason Jr. explains Variation. “That’s the distinction we’re trying to make. The human award that highlights excellence driven by human creativity.”

The full text of the decision is as follows: The GRAMMY Award recognizes creative excellence. Only human creators are eligible to be considered, nominated, or win a GRAMMY Award for a GRAMMY Award. A work without human authority is not eligible in any Category. A work containing elements of AI material (ie material produced using AI technology) is eligible in the applicable Categories; however: (1) the human authority component of the submitted work must be significant and more than trivial; (2) the human authorship component in question must be relevant to the Category in which the work in question is entered (for example, music and/or lyrics if the work is submitted in a lyric Category; if the work is submitted in a performance Category, it must be more than performance-wise and de minimis human authorship ); and (3) the author(s) of any AI material included in the study are not eligible to be candidates or GRAMMY recipients so long as their contribution to the work consists of such AI material. De minimis is defined as devoid of importance or significance; small enough to deserve to be ignored.

The highly detailed statements reflect confusion about AI and the creative arts in general; Paul McCartney was boldly relieved last month when he revealed that AI technology was used to clean up a rough recording of a late ’70s John Lennon demo. create a “new” Beatles song. As alarmists create visions of a purely AI Lennon string and voice – something like this: 100% AI-generated voices of Drake and the Weeknd in the song “Heart of My Sleeve” The technology, found to be copyright infringing and removed from streaming services, was instead used to sonically clean the recording, possibly removing hissing and background noise and improving audio. McCartney later explained, “Nothing was created artificially.”

Mason stressed that the last-minute rule review was done to clarify the fact that this type of AI technology is good, but creative It’s not AI – where an AI program, for example, takes multiple samples of Drake’s songs and then recombines them to essentially create his own Drake song.

“We wanted to say that materials using artificial intelligence could be sent,” Mason said. Variation“but the human part of the composition or performance is the only part that can be awarded or considered for a Grammy Award. Thus, if an AI modeling system or application creates a piece – if it “wrote” lyrics and a melody, it would not be eligible for a composing award. However if a human writes a piece and artificial intelligence is used to model a sound or create a new sound, or uses someone else’s voice, yield wouldn’t be appropriate, but the writing of the track and the lyric or top line would definitely be eligible for an award.

Central to the phrase is the legal term “di minimus”. Cornell University Law School You describe it as “something very insignificant or unimportant”. Usually in dollars it refers to something so minor in importance or seriousness that the law ignores it. To be eligible, the human contribution to the song must be greater than “di minimus”; This means that even if a song has a purely AI-generated “feature”, it may still be appropriate.

In fact, he confirmed that this meant the “new” Beatles would likely be available.

“The provision states that as long as there is more than minimal human participation in the part of the creativity that is being considered for nomination, it will still be considered for a nomination,” says Mason. “So, if you have a rap record with eight measures of AI rap, but the rest of the song is human rap and it has a human chorus, that would still be eligible for a performance award. There wouldn’t be a reward for the AI-generated track, of course.

“When we talk to Paul McCartney about this Beatles record, I think it’s a good comparison: if three or four Beatles are singing on the record and one of the voices [has been sonically enhanced by] AI is still a lively human performance with a very low amount of Beatles, so it would be appropriate.

In fact, Mason said he talked to “Ghostwriter”, the eponymous creator of the fake Drake-Weeknd duet.

“I had the chance to talk to ‘Ghostwriter’ and listen to his perspective, and it’s very interesting,” says Mason. “He’s very forward-thinking, very creative. And from my point of view, it’s been an exercise for him to have a dialogue and raise some awareness of the possibilities and some of the potholes. I hate smacking, but I think he totally understands what he’s doing – that He knew it would be controversial.

Earlier this year, the singer-songwriter Grimes said he has no objections to people using AI-generated versions of his voice. He even developed an app to help people do just that, as long as he shared half of the resulting song’s royalties with him. At the National Music Publishers Association meeting last month, Madonna executive Guy Oseary said they were much more interested when he talked to artists who vehemently opposed AI about a revenue-sharing AI model.

“I haven’t spoken to Grimes,” Mason said, “but the Grimes model seems like a viable avenue for artists who want to let people use their voices and likenesses. The financial model and copyright structure around what he’s establishing makes sense in theory – sure, a bit of tweaking and It should be a bit of nuanced development, but it seems like a very plausible possibility. And you’ve identified two critical issues: credit and pricing. And if we can figure these two things out, the approval process is no different than the normal approval process when you’re trying to use an artist’s song or an artist’s picture or other uses.

Stating that he had preliminary meetings with the US Copyright Office about AI and interacted with Senator Majority Leader Chuck Schumer’s team, Mason added, from his many discussions on the subject, both within the Academy and the industry in general: It is a general feeling of optimism. [rights-holders’] side. Instead of trying to forbid, hinder, or prevent, we approach it in terms of what the possibilities might be, because you can’t. That’s why we need to come together and solve this problem.”

You may also like

Leave a Comment