The quality of Machine Translation (i.e. translation performed without direct human intervention) keeps improving, and yet it is often unfavourably considered. Machines don’t look like us humans, obviously; moreover, machines don’t think. How could they be able to translate? Yet translate they do, even though automation and thinking are often seen as complete opposites. Work on thought systematisation and automation – applied to language – started a long time ago. Kircher, Wilkins, Leibniz and a number of others sought universal harmony, quite often as a remedy for the Babel “Disaster”. They developed new languages that would be free of any defects or translation mechanisms that anybody could use. The methods they used sometimes bear uncanny resemblance with current Machine Translation processes. Alan Turing, who worked on automatism as a concept, played a pioneering role. There is therefore a clear case to be made for reconsidering some of our biases and abandon the comfort of obsolete certitude.