"the gap between the underlying population and those who design the blueprints for rationalization, who lay out production, who make the inventions and discoveries which accelerate technological progress, becomes daily more conspicuous." (Marcuse 1941(Marcuse /1982 Abstract: Recent research on translation memories and machine translation technologies tends to focus on technical issues only, falsely abstracting the technologies from the many different social situations in which they are ostensibly to be used. At the same time, the revolutionary promise of the systems with learning potential is that they will improve output only with widespread use, and thus only through the involvement of different groups of social users. In principle, humanistic research is well positioned to investigate and communicate between the various users, with awareness of different kinds of social actors, collaborative workflows, text types, and translation purposes. If knowledge on those variables can be fed back into the technical research and development, humanistic research could play a key role in enhancing not only the social impact of the technologies, but also their democratization.Translation is increasingly carried out using translation memory systems (TM) that incorporate machine translation (MT), thus giving TM/MT systems. The MT component in most contemporary systems is moreover statistical or data-based, in addition to various more narrowly linguistic algorithms. These technologies can be associated with many changes in the way people produce and use translations. For the perspective adopted in this paper, however, the most important features are the following:1. In principle, the more you use data-based TM/MT, the better the output delivered by the system. This is what we are calling the "learning" dimension. 2. In principle, the greater the online accessibility of TM/MT systems ("in the cloud" or on data bases external to the user), the greater the number of potential users and the wider the range of users.These two features are clearly related in that the greater the public accessibility, the greater the potential use, and the greater the likelihood of improved performance. In short, these features support each other and could constitute something like a revolution,