Developing criteria for algorithmic systems designed in the public interest
The increasing use of algorithmic systems confronts society with opportunities, but also risks. We are convinced that the design of algorithmic systems must therefore be governed by a certain framework. Together with the Bertelsmann Stiftung (Bertelsmann Foundation), we have set ourselves the task of developing one.
Every day, algorithmic systems make decisions for and about us, for example by filtering job applications and diagnosing illnesses, analyzing creditworthiness or suggesting patrols to the police. In other words, they are increasingly used in areas that have a direct or indirect, but always significant influence on society or the lives of people in it — one that goes far beyond the reach of mere pocket calculators. This is why we as a society must discuss how the opportunities offered to all of us by use of this technology can be made to outweigh the risks.
Algorithmic systems are not only designed by programmers; many others, from executives to users, have a role in shaping them. Therefore, we especially value a participatory approach, with representatives from science and research, from companies and associations, civil society and NGOs, as well as from politics and administration.
After an open, participatory and interdisciplinary process (in which more than 400 people have taken part), numerous workshops and consultations, and a large survey, nine so-called Algo.rules were formulated. These are a catalogue of formal criteria that, if adhered to, are designed to facilitate the socially beneficial design and testing of algorithmic systems. Our purpose is to lay a foundation for ethical considerations as well as for the implementation and enforcement of legal standards in this field. We are currently working on putting the Algo.rules in concrete terms specific to various target groups. Further information is available on the Algo.Rules website.
Telefon: +49 30 40 36 77 230
Fax: +49 30 40 36 77 260