Enter the metrics: critical theory and organizational operationalization of AI ethics

Abstract As artificial intelligence (AI) deployment is growing exponentially, questions have been raised whether the developed AI ethics discourse apt to address currently pressing in field. Building on critical theory, this article aims expand scope of by arguing that addition ethical principles and design, organizational dimension (i.e. background assumptions values influencing design processes) plays a pivotal role operationalization development contexts. Through prism notions underdetermination technical code as Feenberg particular, related two general challenges operationalizing AI: (a) challenge placing conflicting demands an cannot be satisfied simultaneously, for which term ‘inter-principle tension’ coined, (b) translating principle technological form, constraint or demand, ‘intra-principle coined. Rather than discussing principles, methods metrics, notion precipitates discussion subsequent value decisions, governance procedural checks balances. It held including interrogating context approaches allows more depth understanding current concerning formalization implementation well ways these could met.