Governance in the era of data-driven decision-making algorithms

Massive streams of human behavioural data, combined with increased technical and analytical capabilities (in particular, data-driven machine-learning methods), are enabling today’s companies, governments and other public sector actors to use datadriven machine learning-based algorithms to tackle complex policy problems (Willson 2016). Decisions with both individual and collective impact that were previously taken by humans – often experts – are nowadays taken by data-driven artificial intelligence systems (i.e. algorithms), including decisions regarding the hiring of people, the granting of credits and loans, judicial judgements, policing, resource allocation, medical diagnoses and treatments, and the purchase/sale of shares in the stock market. Data-driven algorithms have the potential to improve our decision making. History has shown that human decisions are not perfect – they are subject to conflicts of interest, corruption, selfishness/greed and cognitive biases, which result in unfair and/or inefficient processes and outcomes (Fiske 1998). The interest in the use of algorithms can therefore be seen as the result of a demand for greater objectivity in decision making and for a better understanding of our individual and collective behaviours and needs.

Sustainable Development Goals:

Gender Equality

Reduced Inequalities

Partnerships for the Goals