Today theusage of Big Data Analytics and Artificial Intelligence in practice replaces processes ofhuman decision making more and more. Currently, in the context of welfare institutions there ismarginal exploration of usage of Big Data yet. The aim of the project„Machine Decision Support inWelfare institutions: Usage Options, Implications, and Regulatory Needs (MAEWIN)“ is both todevelop a prototype of machine decision support for the use in Social Work, and to give an overviewof Ethical, Legal, and Social Implications (ELSI). During the process of development we have to makesure to safeguard democratic principles from capriciousness, and to prevent the loss of transparencyand comprehensible criteria of decision making. In our speech we will present the proceed of theproject. First, we will explain the results of the field exploration related to layers and situations ofprofessional decision making in this context: Which kind of data are available? Which data are used(and not used) in the process of decisionmaking? Second, we will show, how to handle theCollingridge dilemma. This dilemma describes the problem to combine both in a earlystate, theprocess of planning.