Bias II for piano and interactive music system is part of a series of works engaging with the materiality and limitations of Machine Learning (ML) algorithms and data. During its interactions with different pianists, the computer music system collects data pertaining to the way performers navigate a set of 7 timbral clusters (pools of timbrally similar musical actions). Each second of the performance, a Recurrent Neural Network (RNN) trained on these data predicts how the performance might continue (i.e., which timbre is likely to follow next) and plays back sound material based on its predictions. Through this ML process, the work sets performers in an explicit dialogue with its interpretative history (i.e. interpretative choices made by themselves and other pianists in past performances). 

Historical data, collected by the computer music system in past performances, influence the system’s future behavior, establishing a reciprocal relationship between individual performances of the work and its inscriptions (data). Rather than being an independent, self-contained event, each performance of this piece is a link in a chain of co-creative acts that both instantiate and rewrite the work. Bias II explores distributed musical creativity—distributed among actors (composer, performers, computer music system) in space and time, and reframes musical authorship in collective and posthuman terms.