Data and Algorithm Ethics

Status: 
ongoing
Period: 
October 2017 - October 2020
Funding: 
69500 €
Funding organization: 

Fondazione Bruno Kessler

Person(s) in charge: 
Executive summary: 

The main goal of this research is to investigate the impact of data revolution in our society, combining data analytics techniques with an epistemological approach from social sciences and humanities.

Background: 

The availability of large-scale data, especially on human behaviour, is profoundly changing the world in which we live, putting society in a position of very strong transition.

The flow and subsequent analysis, often automated, of this type of data is offering a huge opportunity for the academic and non-academic community to understand human behaviour and help decision-makers make better decisions on issues of social importance. Thus, researchers, companies, governments, financial institutions and non-governmental organizations are actively experimenting algorithmic tools for decision-making, often based on the analysis of personal information.
However, researchers from different disciplinary backgrounds have identified a number of social, ethical and legal issues related to evidence-based decision-making, including confidentiality, transparency and accountability, bias and discrimination. For example, the use of evidence-based decision-making processes can lead to disproportionate negative outcomes for disadvantaged groups, in ways that resemble discrimination.

Objectives: 

Research will identify ethical issues raised by the collection and analysis of large amounts of data, such as the possible re-identification of individuals as a result of the use of personal data and the risks of group discrimination (e.g. age, ethnicity, sexism).
It will also focus on the problems arising from the increasing complexity and autonomy of some algorithms, especially in the case of machine learning applications; in this case the focus will be on unexpected and unintended consequences and on the responsibilities of designers and data scientists.

Results: 

The first result of this research is a short paper on data labels, which are proposed to avoid future problems raised by incorrect use of statistical sampling methods.