In this chapter, an asynchronous translation exchange via document sharing between American and Japanese university students involving a three-phase process is examined through qualitative data and Rasch analysis of learner beliefs about L2 translation and foreign language learning. The purpose of the translation exercise was for learners to see the text as a unified whole, while also viewing it at the sentence, clause and word levels. This was addressed in real time with classmates and contemporaneously with their peers overseas. The synchronic stages of the project include: (1) draft translations of source texts, an excerpt of a movie script of 295 words for the students of Japanese, a Japanese newspaper article of 514 characters for the students of English; (2) feedback from peer counterparts in the exchange country, a copy of the source text translated through machine translation, explanation and guidance from the instructor; and (3) revision of the draft translation based upon feedback. Results indicate student interest in the use of AI in their coursework albeit with uncertainty about how best to employ it. Peer feedback was found to be more trustworthy by participants due to the human connection and perceptions of their own limits in evaluating AI output.