As a composer, Tod Machover has received commissions from some of the world’s leading orchestras and ensembles. As an inventor he’s currently the director of the MIT Media Lab, and leads the Opera of the Future group, which has spun out innovative music technology such as Guitar Hero. He spoke with Cambridge Day’s Jason Zhang about his new work “already and not yet,” an interactive concerto commissioned by the Boston Symphony Orchestra (BSO) and performed by the BSO and a generative AI model Machover’s lab developed.
This interview has been edited for clarity and condensed.
Cambridge Day: Tell us about “ already and not yet.”
Tod Machover: The BSO got in touch a couple of years ago and asked if I wanted to write a piece. I said that I would be really interested in writing a piece without technology (which ultimately did not happen). I really wanted to write something about society and democracy, what was at risk, and what could possibly happen. And I thought of the orchestra as a society on stage that could tell an interesting story about individuals and groups, who’s in charge, and how you negotiate where to go.
What inspired it?
We (the Opera of the Future lab) have been developing various kinds of AI in our music for a long time. With the new generation of systems, we did a piece called Flow Symphony, we built a specific model with music I wrote and sounds recorded from a river in Vermont, and we tried to make fits between natural sounds and string sounds. Right now, I find that the AI models are particularly useful for making sensitive and delicate variations on a single line of music. They’re also pretty good at making mash-ups or combinations between things which don’t necessarily seem to fit. Both of those ideas are very resonant with the idea of individuals versus communities. And so I decided there would be an electronic layer with a lot of things that I make in my studio, and then a live AI layer. The subtitle of the piece is “Concerto for Orchestra with Electronics.”
And what do you hope audiences will take away from the experience?
I wanted to have three movements, “Smash Up,” “Wake Up” and “Stand Up.” The first movement is about the shock of having forces all of a sudden appear that are trying to kind of stop everything you care about. The orchestra’s organized in normal sectional blocks, although the strings and winds are trying to spin out these melodies and counterpoint and the brass and percussion keep trying to shut them down and eventually just scream and explode and everything kind of stops.
The second movement basically treats the whole ensemble as a group of soloists. There are a few melodies that tie the whole piece together, but in the second movement, fragments of these melodies come out from different solo instruments that, little by little, pass it around, and then join in, and there I use the AI system to be sensitive to subtle changes of nuance in playing style between timbre of instruments. We can change what it’s looking at in many different ways while the orchestra is playing. Manaswi Mishra (Machover’s student) will be on stage right in front of the conductor, making sure the model doesn’t crash, but also pointing the system in different ways over the course of the whole piece.
In the third movement, it’s looking for connections between blocks of instruments, rather than individuals. We’ve tuned [the AI] so that it’s especially sensitive to the contrast between things that are the same — everybody playing one chord at the same time — and things that are quite different — everybody playing something different. We have sections where it goes back and forth between these two types of instructions, and the orchestra plays, and the system listens, and then it responds, and then the orchestra responds. The third movement is based on trying to arrive at a different kind of balance.
We’re also putting speakers around the hall. So in that third movement, little by little, the sound does creep out into the hall, and by the end, when they’re trying to make this new balance, the idea is it’s also embracing the audience, and hopefully, something special happens.
It seems like you’re leveraging a lot of these technologies to make the messaging of how democracy ties to the society of the orchestra more clear, arriving at some kind of consensus between collective and individual voices.
Maybe I’ll send you out there to explain the piece! I wanted to add that the whole public perception of technology, and especially AI, has changed so much in recent years. A lot of AI is pretty scary. It’s not transparent; nobody knows what’s going on inside.
Like you said, the public perception of AI is changing, with many people starting to view it as antagonistic to democracy. How did you think about what it means to include AI in your music?
Many days I woke up and said, “God, I should have stuck to my original idea and not had any technology at all.” But then, technology’s kind of part of my life, it’s part of the way I think. I’ve always taken it as part of my responsibility to try to make models of technology that enhance human abilities and enhance human intention and connection. There’s always been a tendency for technology to make people conform to it rather than the opposite, and it’s just so exaggerated right now.
So, the kind of systems we’re using in this piece are meant to be dialogue systems and collaborative systems. They don’t have any models of the BSO inside; they don’t have models of other people’s music inside. They all react live, immediately.
The AI system is going to react differently every time the orchestra plays. The orchestra has the score, it’s written so that the orchestra and individual soloists have the ability to respond to that with a fair amount of freedom. There’s a lot of room for surprise, and for the orchestra to actually negotiate something that feels really fresh. Hopefully it’ll work and feel that way.
I’m particularly excited about the upcoming conversation you’ll have with George Lewis (composer, technologist, and professor at Columbia), because in the 1980s he was developing Voyager, an early live AI improvisation system. In trying to develop the musical AI interactions, did you look to the past, did you look back to your own work, or any other work that’s been done in this area?
I’ve known George since the beginning and we’ve been friends for all these years, and in a certain sense, we’ve taken an opposite approach to interactive systems. Back in 1980, his goal was to develop systems where he put some of his DNA into the software. He really thought of those systems, and still does, as completely independent musicians.
I always viewed these systems as ways that an artist could have extra layers of expressive control over music that was composed. You knew what the shape of things were, but there were all kinds of things you might want to bring out in a performance, or the kind of performance variety that really interested me.
And I think there’s something between his approach and my approach, which is where I want to go next. I’m not saying that’s where the field should go, but that’s kind of what I see.
“already and not yet” was premiered by the Boston Symphony Orchestra Thursday and will be performed Friday and Saturday.
A stronger
Please consider making a financial contribution to maintain, expand and improve Cambridge Day.
We are now a 501(c)(3) nonprofit and all donations are tax deductible.
Please consider a recurring contribution.