Legal Protection Debt in the ML Pipeline

Organizers

• FARI AI for the Common Good Institute (VUB)

Event Contact

• Carl Mörch (ULB)

Programme

Time	Speaker	Description
13.30 - 15.00	Mireille Hildebrandt (moderator), Gianmarco Gori (introducing the Report as author), Paul Lukowicz (coordinator HAI Net), Masha Medvedeva (CS), Tom Lenaerts (CS), Johannes Textor (CS), Irena Kamera (Law), Michael Veale (CS, STS, Law)

Background

The machine learning pipeline kicks off with the collection and curation of training data. Besides the environmental impact of generating, storing and processing ever more data, there are more reasons to question the sustainability of this type of data-driven AI systems. This panel will engage with a report written by Gianmarco Gori for the Human AI Network that is focused on groundbreaking research into human-centred AI. The report investigates how a ‘legal protection debt’ builds up in the ML pipeline where upstream design decisions may have a major downstream impact on the fundamental rights and freedoms of natural persons. The report confronts the legal duties of the GDPR and the legal framework for open data and open science, highlighting the need for responsible AI from the perspective of legal protection by design. The panel hosts computer scientists, developers and legal scholars, promising an animated debate about arduous issues in data-driven research.

Please consult the report in this link: https://www.cohubicol.com/assets/uploads/hai-net-report.pdf