Stanley Greenstein will defend his doctoral dissertation “Predictive Modelling in a Legal Context” at the Faculty of Law, Stockholm University on June 1st, 2017.
Individuals are spending an ever-increasing portion of their lives within the on-line environment, the internet and social media catalysts in this regard. The by-product of this hyper-connectivity is large amounts of both content and meta data, that hold untapped knowledge concerning the personality and behavioural characteristics of these individuals.
Predictive modelling is a technology based on applied statistics, mathematics, machine learning and artificial intelligence. In analysing examples (historical data), algorithms are able to identify correlations between these data that are invisible to human beings. This knowledge is subsequently incorporated into computer models, where algorithms are tasked with applying this extracted knowledge on novel circumstances, sometimes automatically and in real time. One use of this technology is to identify and predict human behaviour. This insight is a powerful tool, as the ability to predict human behaviour brings with it the ability to influence human behaviour.
Private corporations are increasingly resorting to predictive models in order to identify business risks and vulnerabilities. Predictive models are becoming an indispensable part of the corporate decision-making apparatus, where individuals are no longer judged at face value but rather according to their digital image. This is an essential tool for corporations, as the traditional relationship between corporation and client has been disrupted by human mobility and globalization. As a result, predictive models are determining a large amount of human interaction with the internet: they determine the music people listen to, the news feeds people receive, the content people see, the partners that are suggested, the jobs sought, the diagnosis of disease and whether people will be granted credit, to mention but a few.
This dissertation examines predictive modelling from the legal perspective. This technology is a powerful tool and for all the many associated societal benefits, there are accompanying dangers. This study begins by examining the European legal sphere while inspiration is drawn from the international context. It examines the technology of predictive modelling, making an inventory of the potential harms, a common denominator in this regard the threat to personal autonomy. Thereafter, it investigates the extent to which the European legal regime of data privacy (data protection and human rights) addresses these harms. Finally, a strategy entitled ‘empowerment’ is presented. It comprises an inventory of traditional legal, soft law and technical mechanisms, which, if placed at the disposal of individuals (or possibly taken into consideration by the legislator), will level the playing field between corporation and individual, ultimately protecting personal autonomy by reinstating the power balance.
Institutet för rättsinformatik 