AI as a Creator and Partner: A Stevens Jazz Project With Far Broader Implications
Software that improvises live jazz with human players on the fly, in the style of the all-time masters? A Stevens dean and his team are designing it.
Creating computers and learning models that are more humanistic is a goal of everyone from marketers to device manufacturers to national defense officials. As researchers worldwide rush to design artificial intelligence systems that anticipate our needs, ask us the right questions, or guess how we’re feeling, one Stevens researcher is driving this kind of innovation forward.
He also happens to be the head of the university’s arts and humanities college.
Kelland Thomas, a saxophonist and computer scientist who joined the university’s College of Arts and Letters as dean in 2016, is teaching computers to love jazz by studying the patterns of the great masters. While musically interesting, the research is much more than an artistic venture: if successful, it also has wide implications for defense, industry and an increasingly digital society overall.
The U.S. Department of Defense’s DARPA (Defense Advanced Research Projects Agency) provided $2 million in funding for the Musical Improvising Collaborative Agent (MUSICA) project; the same research powered a collaborative team that has advanced to the third round, so far, in IBM Watson’s AI XPRIZE competition with an artificial intelligence system known as GURU.
“We want to humanize artificial intelligence and show the public that AI can be a powerful positive tool for good, for making things, and doing things together,” explains Thomas.
AI systems can already write songs, short stories, poems and even movies, but creating work with a recognizably human touch remains elusive.
“The goal is for each program to work collaboratively with a human, take turns with a human. Interacting with a real person makes creative products,” says Thomas
With research professor Donya Quick, Thomas builds software that analyzes musical data at scale to mine patterns, create a database and construct probabilistic models. Thus prepared, the system begins to exhibit creativity and improvisation in real time, predicting and reacting to live musicians who are, themselves, improvising new music.
“This is classic AI, classic knowledge engineering,” notes Thomas. “The system’s ability to anticipate and create chord changes and new, never-tried melodic riffs will go far beyond current computers’ ability.”
The Holy Grail, known as computational creativity, will require AI systems to build knowledge, learn on the fly and interact with people in ways that are cognitively similar to the ways we interact with them. It could enable smoother, much more productive interactions with devices, automobiles, homes and intelligent services such as those that power virtual assistants like Siri. And Thomas’ work could help achieve that goal.
To think that it all began with the work of jazz masters.