With the improvement of free, next-generation machine translation software based on neural networks, Departments for Translation Studies are faced with the question of how to react to this new reality. Translator education programmes should include these and other tools that can complement human translation. However, in order to design activities that incorporate this software, we need data and experience that can serve as a starting point. For this purpose, we have conducted an experiment in a General Translation class and a Specialised Translation class. This article shows the results of this research, including both the quality of the translated text and the students' impressions.