This lecture examines how machine or artificial intelligence (AI) translation mediates a wide range of high-risk intercultural encounters. Analyses of these encounters are vulnerable to several pitfalls. Polarised discourses on AI often overstate what translation technologies can do. At the other end of the spectrum, these discourses may also reject these technologies outright, irrespective of their benefits. This talk charts a different path. Its focus is on the experiences of UK critical professionals, including doctors, paramedics and police officers. The talk looks at these professionals not to judge the adequacy of their decisions but rather to examine how their work environments and the affordances of AI give rise to complex communication practices that merit attention from language researchers. The discussion will draw on a range of sources, including legal decisions, official documents and findings from Translate: Multilingual Access to Critical Services in the Age of AI. The talk will defend the value of nuance. It will support a broadening out of conceptions of translation and of the social roles fulfilled by translation and interpreting specialists.