TABOO - TRANSGRESSION - TRANSCENDENCE

Biologists have long criticized the oversimplification in the analogies. In the development of modern neural networks, noting that real neurons are vastly more complex than artificial nodes, and that brains do not learn via backpropagation the way our algorithms do. Instead of seeing this disparity as a failure, I argue it is the point: like a translator capturing a poem’s essence over exact wording, bio-inspired algorithms carries the original meaning across when they embrace interpretation over imitation.

This talk emerges from my experience in translator and from my occasional encounters with biologically inspired algorithms during my study in computer science. Repurposing parameters in translation theory as the meta-framework, the mapping between biological mechanism and computer algorithm are dissected in dialogue. In particular, five central ideas, supermemes, from translation theory, is utilized to examine how knowledge moves and transforms between biology and computing: source-target, equivalence, untranslatability, free versus literal, and the notion that all writing is translation. By representing the exchange between organism and machine as a translation, a new vision may be conjured, one that understand bio-inspired algorithms as free translations across epistemic boundaries.