Machines Unlearning: Toward an Allagmatic Arts of AI Available to Purchase

Machine unlearning is a small AI subbranch investigating ways to “unlearn” data points previously learned by a model. In this final chapter, it becomes a broader project for an artful retracing of the operations of machine learning models in a bid to undo claims on prediction and determinacy. This “retracing” is conceived as an “allagmatics,” borrowing from Gilbert Simondon. The chapter examines artful techniques for following the operations of ML that also analogically enact its operations, with a twist. This allagmatic art engenders difference between retracing and enacting, leaving open a margin of indeterminacy. In the work of Anna Ridler, Philipp Schmitt, Tega Brain, and others, a different sensibility and deepaesthetics begin to register for AI.