This article proposes that machine and human translation differ in use and seeks to demonstrate this not by comparison, but rather using the framework from philosophy of technology known as postphenomenology. Following the work of Don Ihde and others, postphenomenology is intended to examine how technologies mediate our relations with the world. Following a short introduction to postphenomenology, the place of machine translation (MT) along Ihde’s continuum of interaction types is considered, and a set of predefined dimensions are applied in order to analyse how MT mediates communication. MT is found to be hugely useful, but problematic in developers’ use of training data and in how MT is described and marketed. The framing of MT contributes to heightened quality expectations and translators’ alienation. The broader alienation of those disenfranchised from ICT use, who are also most affected by carbon emissions and environmental pollution, contrasts starkly with the positive effects of MT.