Machine translation, and specifically Google Translate, is becoming a favourite tool for language learners. As it is freely available across a variety of platforms for both desktop and mobile devices, students tend to rely on it excessively for solving their course assignments, which is ultimately detrimental to their language learning. However, the functionalities of machine translation can be exploited proactively within the curriculum and activities involving its use can be included in a course in order to increase language awareness and ultimately language learning. This paper is a contribution to the research field of machine translation in language pedagogy and reports the results of an empirical study conducted on students of an Arabic beginners course at a Swedish University. The study aimed at making the most of the students’ attitude of relying on Google Translate, while at the same time taking advantage of GT’s capabilities and limitations as a tool for language learning. A brief introduction covers machine translation and the reasons for GT inconsistencies with translating Arabic. This paper then presents two assignments given to the students at two separate moments of the term, accounting for their scope and execution. An analysis of the results follows, highlighting the students’ performance and reflections on the advantages and disadvantages of GT.