Arbitrary Intelligence investigates new musical forms proposed by algorithmic systems of intelligence and develops strategies for engaging with the alternative musicalities articulated by these systems. It questions the anthropocentric notion of intelligence and instead explores how diverse machine-learning architectures embody alternative forms of intelligence. While such systems may exhibit behaviours recognisable as humanly-intelligent, algorithmic reasoning structures offer vast compositional possibilities that can give rise to coherent yet seemingly arbitrary musical organisations. This research therefore addresses not only the new musical possibilities enabled by their structural affordances, but also the question of how musical coherence can be established through compositional and presentational engagement strategies. It mediates the tension in computational creativity between opaque self-referential systems and mimetic human-aligned systems. Situated within the context of human-machine co-creativity, the project explores alternative forms of intelligence that introduce new modes of musical organisation and extend habitual notions of musical coherence.
