It’s Here: AI in Jazz Education

By Yoshi Kato I Sep. 18, 2025

“It’s happening very quickly,” says Berklee’s Michele Darling of AI in the classroom.
(Photo: by Kelly Davidson)
“I still stick by the idea that the really great music is going to come out of people …” says Princeton’s Rudresh Mahanthappa.
(Photo: By Renato Nunes)
“I’m nowhere near a coder. But I’ve built an AI system on my laptop over the years,” says Berklee’s Rodney Alejandro.
(Photo: Courtesy Rodney Alejandro)
“We have to start conceiving of it as a thought partner that can come to some understanding of the way that you think and communicate,” says Rugters’ Stefon Harris of AI.
(Photo: Shahar Azran for The Apollo)
“I’ve told students, ‘I don’t mind if you use AI for this or that project,’” says MIT’s Pascal Le Boeuf. “‘But you need to tell me.’”
(Photo: Nadine Sherman)

A standard joke when it comes to discussing artificial intelligence, or AI, is that it’s developing so rapidly that the information contained in this article will be woefully out of date by the time it’s published. And while a computer’s ability to complete and now sometimes surpass tasks done by humans is growing exponentially thanks to technological advances (and natural resources), the unprecedented approach to including it in jazz education has remained the same so far.

Michele Darling, Berklee College of Music’s first-ever assistant chair of the Electronic Production and Design Department, is an ideal guide into the world of AI and music education. She laid out the three major areas of AI that are currently in the ether.

Generative AI takes prompts, or user inputs, to create new works such as a realistic picture of polar bear surfing in Hawaii or a ragtime piano roll recording of the latest Taylor Swift album. Assistive AI aids tasks like editing an essay or creating a wedding reception menu centered around locally sourced ingredients. And Analytic AI crunches massive amounts of data for use in fields such as medicine and economics.

Jazz education is likely to utilize the generative and assistive versions of AI. “It’s happening very quickly,” Darling confirmed, when asked about the technologies’ development and its implementation in the classroom. In the very first semester of instruction, the groundwork around AI is established.

“We talk about what’s important ethically and morally and what’s best for learning. We try to teach how to learn, because critical thinking is key around all of this,” she said. “We ask them, ‘Why are you using AI? What is the result of AI? What effect is it having on the planet or on your learning? Or on plagiarism? And is it taking from someone else’s artistry?’

“First and foremost, we’re still helping a human being develop here,” Darling replied when asked about educators’ relationships to their students in the burgeoning age of AI. “This is a technology, a tool. How do you develop a person to be their most creative self, to understand who they are as an artist and to make sure they have the fundamentals of music and of the technology, if they’re interested? And, while doing this, we do have to teach the tools of our time and their time.”

Detailed in DownBeat’s May 2024 cover story, vibraphonist and educator Stefon Harris has been incorporating AI into his performances. As both an associate professor of music at Rutgers University and artistic advisor of Jazz Education at the New Jersey Performing Arts Center (NJPAC), he’s been utilizing AI educationally, as well. “AI still hasn’t quite mastered the space around music training,” he said. “But it can certainly talk to you a lot about music theory.”

Harris’ own Harmony Cloud iOS app, for instance, uses AI for better aural recognition. “It provides a harmonic backdrop for students,” he said. “Harmony can be very, very complicated, and one of the things that we often overlook in education is a strong emphasis on melody making. So many of my classes primarily focus on melodic invention, and Harmony Cloud actually takes care of creating melodies so you don’t have to worry about that initially.

“We have to understand our relationship with AI,” Harris went on to advise. “We often think of it as a tool. But I think we really have to start conceiving of it as a thought partner that can come to some understanding of the way that you think and communicate. Even as a musician, I use AI often in my own study, just asking questions about how to articulate some of my ideas.”

“We’re experimenting with AI as a tool for composition,” said Rodney Alejandro, dean of professional writing and music technology at Berklee. “We have a class in songwriting called Bots and Beats, and it’s basically how to work with AI bots to create (rhythmic tracks) to work with lyrics.

“What we try to avoid is the temptation for students to use AI just to create everything,” he said. “We want them just to use AI as a tool to help students get unstuck or discover new ideas or find a new approach to their own work.”

Since each AI platform is constructed on separate data sets, using more than one gives students different perspectives. (And that’s more accessible at an educational institution with its greater resources.) It’s not unlike sitting in a room to work off of someone who specializes in traditional jazz and another musician who’s an expert in Afro-Cuban styles and a third who’s an R&B veteran.

And just as some students thrived in remote learning situations that arose from the pandemic lockdown, so too have some been able to engage with AI before branching out to peer-to-peer collaboration. “Songwriting is about vulnerability, and that can be uncomfortable for some,” Alejandro said. “Those folks can build up some resilience around AI before entering those student circles.”

With commercial AI services based on massive libraries of external input, Harris and Alejandro both encourage students to build their own personal AI. “I’m nowhere near a coder,” Alejandro admitted. “But I’ve built an AI system on my laptop over the years.”

“There’s nothing else on Earth that sounds like my AI, because I’ve trained it specifically with my information that I have not made available to open AI sources,” Harris said.

One issue that’s already arisen with AI and education is cheating. “A lot of schools have required statements that they put in their syllabi about needing to disclose use of AI,” reported Pascal Le Boeuf, the new assistant professor of music composition at M.I.T. “I’ve told students, ‘I don’t mind if you use AI for this or that project. But you need to tell me.’”

“Instead of just turning in an end paper or a completed piece of music, they have to turn in their entire session,” said Darling. “You see several iterations of work, so you can see the progress.” Like showing one’s work in math? “Yes, exactly.”

The temptation may be there for students to use AI for heavier lifting when it comes to assignments and spend more time, say, at on-campus or local jam sessions. Rudresh Mahanthappa, Anthony H.P. Lee ’79 director of jazz at Princeton University, is in a unique position to remark on why one would want to avoid that temptation.

“I’m in an interesting situation running a jazz program at Princeton. Very few of the students that I deal with have great intentions of being professional jazz musicians. They’re there to study something else,” he shared. “We’re not a conservatory, and we’re not a music school. We’re a music department, and all of our ensembles are extracurricular.

“That is to say, they’re all there because they love the music. And if they’re taking an arrangement class, they really want to learn the stuff,” he said. “They’re not there because it’s a requirement towards their degree. They’re there purely to acquire this knowledge. So if they’re going to use AI to write their big band charts, it’s their loss.”

AI is a topic of both fear and fascination in 2025. Professions such as language translation, computer programming and ride-hail/taxi driving either have the potential to be severely impacted by AI or are already seeing its effects.

On the music side, the allegedly AI-generated band/“synthetic music project” called The Velvet Sundown was garnering news, airwave minutes and column inches not for its songs but rather its streaming numbers and reported fully digital origin story. With the entire history of recorded and notated jazz to draw from, AI can theoretically create any musical scenario that a user can imagine. Whither the college educated musician?

“It’s not like the software is going to replace the act of a group of humans coming together to enjoy the process of making music,” Harris observed. “These tools put you in a position where you can develop the skills necessary to actually be able to listen and interpret and contribute when you’re in a group of users. But I’m not worried about it replacing the actual creative process.

“It’s similar to sports. People will watch a robot basketball league as a novelty, but it’s not going to replace watching play because you know that you can never do what that robot is doing,” he said. “When you watch the (human) pros, you almost imagine yourself. You see your aspirations in what they’re doing.”

“AI doesn’t necessarily move people. It doesn’t feel the same as when you have real people recording or performing,” Alejandro concurred. “Also, AI is currently only trained on the past. And the one thing that artists do is set the trend in the now and then move towards the future.”

“I still stick by the idea that the really great music is going to come out of people who were already creative and are on their way to developing a unique voice or personality,” Mahanthappa concluded. DB