How Musicians Can (and Should) Use AI-According to Berklee Experts

How Musicians Can (and Should) Use AI-According to Berklee Experts At Berklee's first AI and Music Innovation event, artists and educators explored how artificial intelligence is reshaping creativity and artistic responsibility-and what today's musicians need to know.

By

John Mirisola

May 29, 2025

Illustration by Vanessa Rossi

Its no wonder the first event in Berklees AI and Music Innovation Series was packed to the doors. For the students and faculty in attendance, all current and future music industry leaders, the stakes could not be higher-professionally, ethically, artistically. A year ago a large language model (LLM) couldnt tell you how many Rs are in the word strawberry and were scoring in the 60s on standard IQ tests. Now AI tools for musicians can write you ten versions of a song, generate the audio in ten different styles, master the audio, and write the press release, all with a few plain language prompts.

Its a complicated, mixed bag of truly existential changes-negative threats and positive opportunities, said Berklee President Jim Lucchese, introducing the new speaker series, which will take place monthly beginning this fall. Its really critical for everybody here . . . to all be as well informed in a very fast moving space as possible.

Yet contrary to the fear and pessimism that sometimes overshadows conversations about artificial intelligence and the arts, these 90 minutes were filled with the curiosity and pragmatic optimism of professional artists whove always found new ways to create with cutting edge tools.

For the folks in this room, said Lucchese, its about understanding things well enough to focus on and create those positive opportunities.

The panel, moderated by Chris Wares, interim chair of the Music Business/Management Department, included Ben Camp, associate professor of songwriting; Lori Landay, professor of cultural studies; Jennifer Hruska, assistant chair of the Electronic Production and Design Department; and Jonathan Wyner, professor of music production and engineering. Among the panelists, no single viewpoint dominated. No one-AI-fits-all approach emerged. Instead, each speaker shared their own experiences teaching and creating with artificial intelligence tools, and invited the audience to define their own approach, guided by judgment, ethics, and personal taste.

Here are some of the conversations key takeaways:

Your Taste Is EverythingMuch of the conversation revolved around the importance of artists guiding the creative process when working with AI.

We've crossed a threshold where no creative task feels safely out of AI's reach. Humans are already collaborating-and competing-with machine-generated content across every medium. So what do we carbon-based creators still bring to the table? Well, we know what we want. And we know what we like.

Artistry in the Algorithms: Music in the Age of AI

Berklee alumni and professors speak to Berklee Today about the promise and peril of making music with AI.

Camp described using a large language model to churn through dozens of lyric variations, refining and reshaping them based on their own instincts through generation after generation of prompts. The reason I'm able to navigate these things so quickly is because I know what I want, they said.

Wares illustrated a similar point with a real-world example: an AI might be able to write a decent press release, but only someone with a trained eye and industry insight knows what makes one truly work.

If you don't have that expertise, if you don't have that training, that taste, it may seem like a beautifully written press release, said Wares. But then you hand it over to a publicist who has 10 years of experience . . . they're gonna take a look at it and say, Oh, no. This will never fly.'

Or, as Camp put it: If you dont have the taste to discern whats working and whats not working, youre gonna lose out to the people that do have the taste, because anybody can [use AI to write songs].

AI can generate an infinity of beats-but only you can tell if they slap.

Dont Skip the StepsIf taste should guide our use of AI, then building the skills that form our taste is essential. Dont skip the steps, as Wares said.

This advice was especially relevant to the students in the room, who might be tempted by AIs ease and accessibility to use it instead of developing their own skills. Camp put it bluntly: If you have to learn how to write a song, dont have the GPT write a song for you.

But skill-building isnt just for beginners. Everything is use it or lose it, Landay said. You know what happens when you dont practice your instrument, right? . . . You start to lose it. And its true of everything.

If one thinks, Oh, I can outsource that to ChatGPT, Im just gonna pop this PDF in the LLM, and its going to summarize it for me, youre going to lose something. And its not just summarizing. Its these critical thinking and analyzing skills, and thats what it means to be human.

If you dont have the taste to discern whats working and whats not working, youre gonna lose out to the people that do have the taste.

- Ben Camp, Associate Professor, Songwriting

And while not every human skill is worth keeping sharp, we need to be thoughtful about what we trade away, and what we build in its place.

I used to be really proud of myself that I was able to remember phone numbers. I dont anymore. Do I need to be that? No, said Wyner. But in order to move forward, what do you need to be practicing? Theres a great deal of pleasure in letting go of things we dont need to be doing anymore. But ultimately thats got to be replaced . . . by something that is useful to you.

Refining, Not Replacing, Your Creative ProcessWyner has seen AI tools dramatically expand whats possible in audio engineering: I could do things using AI I never could do bef

LINK:	https://www.berklee.edu/berklee-now/news/how-musicians-can-and-should-...
	See more stories from berklee

More from Berklee