Elisa Orrù

Max Planck Institute for the Study of Crime, Security and Law

"From Big to Small Data? Assessing the ethical implications of a potential sociotechnical shift"

"Small Data" is usually used in AI research to point to a problem, or at least a limitation: namely,that in some areas of AI application, such as rare disease detection, processing of rarely spokenlanguages, and so on, the huge amounts of data needed to train AI systems using traditional deeplearning methods are not available. Yet "Small Data" can also be seen as an option and anopportunity: Moving away from the currently dominant Big Data mindset towards a Small Dataapproach would help to improve the sustainability of AI systems and could have additional ethicalpositive impacts. In this spirit, and drawing on my previous research on the topic, in the talk I willfirstly explore the meaning of "Small Data", i.e. the way the phrase has been used in AI researchso far and present and discuss "successful" experiments that have voluntarily adopted a Small Dataapproach. Moreover, I will explore the ethical implications and possible benefits of moving awayfrom the dominant Big Data paradigm towards a Small Data approach, including in terms of:

• sustainability (reducing the carbon footprint and conserving resources such as water, landand rare minerals),

• global justice (as the distribution of gains and losses of the current big data approachexacerbates global inequalities and the North-South divide),

• intergenerational justice (by helping to preserve the availability of resources for futuregenerations),• accuracy (assuming that the drawbacks of less available training data can be compensatedby easier curation of the data),

• reduction of bias (as a possible consequence of better data preparation) and

• privacy (because less personal data is processed).

Finally, I would like to discuss with the other participants under which conditions a "Small Dataapproach" can be a technically feasible and preferable option, even if the abundance of data couldallow the continuation of the current (surveillance-capitalism driven) Big Data approach.

Bio

Elisa Orrù is Senior Researcher at the Max Planck Institute for the Study of Crime, Security andLaw and Associate Professor for Philosophy at the Freiburg University. She leads the ethicalcontribution and the Freiburg research unit of the EU-funded project VIGILANT (VitalIntelliGence to Investigate ILlegAl DisiNformaTion). Her research interests lie in the area ofpractical philosophy and focus on the ethics of emerging technologies, digital security practices,privacy, and surveillance. She is furthermore interested in the dialectic between universalism andparticularism of fundamental rights and justice theories, integrating feminist and post-colonialperspectives.