In a new study that digs into the meaning of data ethics in the context of the power dynamics of AI and big data, DataEthics.eu’s co-founder Gry Hasselbalch argues that data ethics has to become more than just a moral obligation and a set of programmed rules—it has to be human.
“One day, 80 years ago, a world-famous philosopher of Jewish descent went to register his data at a police station in Paris due to newly introduced antisemitic laws. Last year in 2020 a man was arrested in front of his house in Detroit for a crime he didn’t commit because of an incorrect data match between his face and the one of a criminal made by a facial recognition system. Data systems have always throughout history reinforced power. But they have also changed form. We are today not only challenged by a database and register of a dominant regime of power”, explains DataEthics.eu’s cofounder Gry Hasselbalch, “we are submerged in socio-technical data systems of power”. This transformation, she says, requires a particular kind of reflection and awareness from us.
Data Ethics of Power
In the PhD study Data Ethics of Power – A Human Approach to Big Data and AI, Hasselbalch presents a framework for a ‘data ethics of power’ looking at our contemporary digital data systems as infrastructures that reinforce and distribute power. She maintains that data ethics has to become more than just a moral obligation, a set of programmed rules—it has to be human:
“We can formulate data ethics guidelines, principles and strategies, and we can even program artificial agents to act according to their rules. However, to ensure a human-centric distribution of power, data ethics must take the form of culture, to become a cultural process, lived and practiced as a way of being in the world.”
With point of departure in Gry Hasselbalch’s immersion in ‘data’ and ‘Artificial Intelligence’ policy, industry and civil society activities of the 2010s in Europe, which among others included the EU’s High-Level Group on AI and the Danish government’s first data ethics expert group, she investigates the role of data ethics and a human (-centric) approach to the ethical governance of AI and big data.
What is Data Ethics?
Based on her many years of participation in the internet governance and digital rights policy, industry and civil society communities, Gry Hasselbalch explores the transformation of a conversation on data ethics that has matured from a small nuisance and pebble in the shoe of big data actors and enthusiasts into a current main topic of public debate and a crucial policy agenda item in Europe and beyond.
She says that data ethics is not only about power—it also is power: “Power for governments, companies, self-proclaimed experts and advisors and even academic disciplines to point out the problems and their solutions, to set the priorities for what role data technologies should play in our human lives and in society.”
Data ethics has become everyone’s declaration, which also means that data ethics as a field of study, as an approach or a concept, is currently in crisis, she argues. Blamed for a lot of things, but most critically, she states, for being a sweet cover for bad data practices of companies and governments. As a form of “ethics washing”.
In this study, Gry Hasselbalch, however wants to set data ethics free. Uprooting the very conceptualisation of the term as the moral obligation of someone or something to solve a specific problem. She does this, as she says, to enable humans to critically challenge the power embedded in data technologies – their set priorities and restraints – and to find different problems and new solutions in the very power conditions of the big data reality we live in.
This is also why she presents a “data ethics of power” as an action-oriented analytical framework concerned with the distribution of power.
The Human (-Centric) Approach
One of the key objectives of the study is to illustrate how the reduction of human complexity is a primus motor for big data and AI policies, business and technology developments and adoption today.
She says: “We want to make life, society and culture easier to handle; and to make those difficult ethical decisions we have to make every day less cumbersome. The ones we make at home, work, school, in our societies in hospital, during elections, in the welfare system, in the justice system and during times of crisis, such as war or pandemics. We don’t want to make these thorny decisions, because we often realise ourselves, or are told by other humans, that we have made them poorly or in ethically problematic ways.”
But this kind of self-conscious critique – that a data process does not have – is exactly why we need to keep making these decisions ourselves, she says and continues: “What I fear the most and have tried to show in my thesis is that a loss of human critical agency is taking form in our sociotechnical realities. The kind of critical ethical human agency that is fundamental to our democratic societies and their institutions. It is basically removed from the very configurations of our architectures and from our imagination, our norms and cultures.”
Gry Hasselbalch points to a recent critique of the “human (-centric) approach” that we for example see spelled out in current EU AI policies and that was promoted in the High Level Expert Group on AI’s ethics guidelines. The critique is that this approach is primarily concerned with the individual human being and the human species as such. That it is anthropocentric. But she thinks we could look at this in an entirely different way.
The human-centric approach she argues is not about prioritizing the individual human being. It is about our human ethical responsibility for not only ourselves but for life and being in general, and it is about prioritising the human dynamic qualities, a human infrastructure of empowerment. This is why the “human approach”, also encourages in practical terms the empowerment of these dynamic human moments in data design, use and implementation, which does indeed also include the empowerment of the individual human being.
The PhD thesis was submitted at University of Copenhagen/Faculty of Humanities January 2021 and defended 13th of April 2021 with a committee consisting of Danish professor Klaus Bruhn Jensen, Dutch law professor Simone Van der Hof and the author of Algorithms of Oppression associate professor Safyia Umoja Noble who is also featured in the current hit Netflix documentary CodedBias. The thesis was supervised by Professor Jens-Erik Mai.
See the PhD defence here
See more about the study here.
Gry Hasselbalch is currently working on the book Data Ethics of Power – A Human Approach to Big Data and AI to be published at the high ranking academic UK publishing agency Edward Elgar Publishing.