Can an instrument determine the behaviour of the performer? Can the free will of an improviser bend to the suggestions of an algorithm? In an age of the rapid development of artificial intelligence, the question of the instrument-artist relations gains in importance. 

To an artist who wishes to feel in charge, the fully autonomous and impenetrable neural network fails to meet the criteria for a usable tool. The “black box,” whose internal mechanisms are unknown, also arouses anxiety among scientists, who therefore aim to create the so-called Interpretable AI. Imperfect interfaces, which create barriers between humans and machines, inspire new implementations of contact with technology. 

AAAA explores the mutual relations between three elements: relatively simple interpretable algorithms, their actuation in the form of hybrid acoustic-digital instruments, and improvising musicians. The feedback loop between these elements generates an artefact that proves highly complex despite the apparent predictability of its components. Changing the parameters of any of these elements will nevertheless still significantly modify the result. 

The instruments I have designed and built allow for digital pitch control, which makes it possible to study a wide spectrum of interactions between musicians and instruments, from free improvisation to looped phrases, to mutually suggesting notes that are to be played, elimination of the most or the least common notes and, finally, a digitally generated piece of music imposed by the instruments on the musicians, who are thus only left in charge of rhythm, dynamics, and articulations. At the same time, the physicality of the acoustic instruments and the nearly tangible quality of their sound reduces the distance between the created sounds, the performers, and the audience. 

Krzysztof Cybulski