As classically-trained rock keyboard wizard Jordan Rudess stepped onto the stage at the MIT Media Lab on 3 October, he engaged in a fascinating, AI-fuelled musical dialogue.
Facing a suite of digital instruments linked to a real-time system dubbed the “Jam-Bot”, the Dream Theater virtuoso was neither playing along to a pre-recorded backing track nor yielding to an autonomous composer. Instead, he was improvising live with a neural network trained specifically on his own musical vocabulary.
Developed alongside researchers from MIT’s Responsive Environments group, Jam-Bot operates in the symbolic domain via MIDI, processing performance data like pitch, velocity, and time shifts in real time (with a sub-50-millisecond delay). By feeding his signature chromatic runs, modal shifts and dense jazz-fusion voicings back to him instantly, the system acts as a real-time “musical mirror”.
Starting around the 16-minute mark of the YouTube livestream (the second of two performances), you can hear this strange synergy in action. Rudess’s talent for spontaneous melodic development is undeniably extraordinary, but the performance highlights a double-edged sword. While the system at times sounds like clever mimicry, listening to Rudess’s solo compositions and neo-prog style can also feel like hearing highly sophisticated and wide-ranging Muzak – admirable in its technical perfection, yet slightly confected, synthetic, and hyper-polished.
Despite the futuristic illusion, Jam-Bot isn’t a conscious partner co-creating on equal footing. Lacking intentionality or awareness of musical beauty, the model is simply executing statistical pattern-matching across a mathematical vector space. It serves as an algorithmic shadow, forcing Rudess to react to his own habits, objectified and fed back to him in the moment.
While commercial AI often seeks to automate human skill, Jam-Bot champions “symbiotic virtuosity” by keeping live human presence and dexterity at the centre. Still, it remains an expensive, highly elite technology. But it naturally begs the abiding question: in an age of automated mirrors, why choose to jam with a machine over living, breathing musicians?
* Jordan Rudess’s web resources can be found here. Screenshots from the second livestream
