IRS 2018 DUTY

On May 25, 2018, the European General Data Protection Regulation (GDPR) will take effect, imposing stricter requirements on organizations that process personal data. The regulation excludes from its scope data that is not associated with an identified or identifiable individual. Thus, at present, this so-called anonymous or anonymized data (AAD) remains exempt from specific legal rules. In addition to this legal vacuum, there is a lack of guidelines regarding data management and processing. Companies can no longer rely on outdated frameworks rendered obsolete by technological and organizational innovations. This results in an increased risk of misuse or even abuse: AAD could harm individuals if used for discriminatory purposes or in an unfair manner.  However, the upcoming implementation of the GDPR will significantly increase the volume of ADAs. Indeed, companies will have to pseudo-anonymize their personal data to comply with the regulation or will likely prefer to process anonymous data. In light of these challenges, ethics—as a reflection on values and behaviors—represents a means of regulating practices associated with ADAs. Through an analysis of these practices, ethics could help identify actions that are desirable—or, conversely, those that should be prohibited—in order to limit the threats posed to citizens. The first step in this process is to bring greater visibility to AI systems and their processing procedures by focusing on the concept of transparency.

The DUTY project aims to study the concept of transparency within the framework of AI ethics in the private sector. By focusing on AI, the project seeks to support the momentum generated by the introduction of the GDPR. From a scientific perspective, data ethics is an emerging field that various disciplines are only just beginning to embrace. Adopting a multidisciplinary approach now would prevent the development of siloed knowledge. This is why the DUTY project involves the joint work of three disciplines: law, management, and computer science. It is structured into three Work Packages (WPs): WP 1 brings together law and management to assess current business practices and applicable legal standards regarding the use of AI as a performance driver. WP 2, at the intersection of law and computer science, aims to determine the discriminatory potential of algorithms and the legal frameworks needed to regulate these tools; WP 3 combines management and computer science to study the implementation of “ethics-by-design.” Each WP is thus carried out by researchers from two different disciplines. This interdisciplinary team can count on the support of the other project members.

Scientific results are expected at the Work Package (WP) level and for the project as a whole. Specifically, the conclusions of each WP will be synthesized to develop, at a broader level, a compliance program that includes: normative rules, an “ethics-by-design” technological tool, and an associated code of ethics. These three elements represent the expected outcome of the multidisciplinary work on the concept of transparency. In this sense, they represent the DUTY project’s contribution to the ethics of AI in the private sector.