Virtuelle Ringvorlesung Wintersemester 2022/2023
Prof. Dr. Christoph Burchard (Professor für Straf- und Strafprozessrecht der Goethe-Universität und Principal Investigator des Forschungsverbunds „Normative Ordnungen“, Clusterinitiative „ConTrust“)
Prof. Dr. Indra Spiecker gen. Döhmann (Professorin für Öffentliches Recht, Umweltrecht, Informationsrecht und Verwaltungswissenschaften und Principal Investigator der Clusterinitiative „ConTrust“)
Online via Zoom. Eine Anmeldung an firstname.lastname@example.org ist erforderlich. Die Logindaten werden nach Anmeldung übermittelt.
Dienstag, 29. November, 18.00 Uhr
Gillian Hadfield, University of Toronto
Judging Facts, Judging Norms: Training Machine Learning Models to Judge Humans Requires a New Approach to Labeling Data
Abstract: Machine learning-based systems developed to make automated decisions about whether rules are violated will not reproduce human judgments if they are built using currently standard techniques for ML data collection.
Montag, 12. Dezember, 18.00 Uhr
Kat Geddes, NYU School of Law
Will You Have Autonomy in the Metaverse?
Abstract: The talk describes how computational prediction is reshaping our tolerance for prediction, and undermining institutional commitments to respect for individual autonomy by increasing the likelihood of ex ante intervention in individual decision-making. Prof. Geddes uses examples from criminal justice (algorithmic predictions of recidivism in post-conviction sentencing) and political democracy to illustrate the autonomy-eroding effects of prediction, and she argues that sophisticated computational models are reshaping the prediction-autonomy trade-off.
Montag, 19. Dezember, 18.00 Uhr THE LECTURE HAS BEEN CANCELLED!
Hannah Bloch-Wehba, Texas A&M University School of Law
Algorithmic Transparency at a Crossroads
Abstract:This talk examines current proposals to promote or require algorithmic transparency. As nations around the world confront disinformation, misinformation, and extremism online, there is a growing consensus that platforms’ algorithms and amplification mechanisms are worsening these dynamics. New efforts to render algorithms “transparent” to policymakers, journalists, and the public represent an attempt to shed light on the obscure inner workings of social media firms. Algorithmic transparency and disclosure requirements are widely viewed as more appealing than substantive regulation of platforms, which is perceived to be more intrusive. At the same time, transparency and disclosure requirements are also a core feature of a wave of increasingly illiberal legislation meant to kneecap social media platforms and compel them to alter their content moderation techniques. This project takes a comparative approach to assess the relative strengths and weaknesses of contemporary algorithmic transparency, auditing, and disclosure requirements. The article critically evaluates whether existing initiatives and proposals are sufficiently dynamic, independent, and rigorous to achieve their stated goals. At the same time, broader algorithmic transparency reforms are underway as states seek to ensure that automated decision systems used in public governance can satisfy due process, equal protection, and open government obligations.
Dienstag, 10. Januar 2023, 20.00 Uhr (!)
Lyria Bennett Mose, UNSW
Mittwoch, 25. Januar 2023, 18.00 Uhr
Katherine Jo Strandburg, NYU School of Law
Justifying Automated Decisionmaking Systems
While explaining the basis of a decision to decision subjects is often important, this presentation argues that other, less often discussed, information flows, such as explanations and communications between other decision system actors, are at least important in justifying the use of automated decision tools. It will also describe early stage work about the potential for agent-based modeling to play a role in decision system accountability.
Mittwoch, 1. Februar 2023, 18.00 Uhr
Johann Laux, Oxford Internet Institute
The New Laws of Oversight: What to Do with Human Discretion in the Age of Artificial Intelligence?
Human oversight over algorithms has become a key regulatory tool for preventing risks associated with automated decision-making. Keeping humans in the decisional loop allegedly maintains some final human control over algorithmic systems. Human overseers are supposed to increase the accuracy, legitimacy, and fairness of automated decisions. However, merely placing a human agent in a hybrid human/machine system will hardly guarantee accurate, fair, or non-discriminatory outcomes. At the same time, recent research shows that people tend to trust an automated system more if a human being is involved in its decision-making. This trust would be misplaced if the human overseer turns out to have little to no influence on the system’s outputs. This talk addresses this regulatory challenge and asks: How can we implement effective human oversight of automated decision-making?
Donnerstag, 2. Februar 2023, 18.00 Uhr
Hannah Ruschemeier, FernUniversität Hagen
Datafication and Collectiveness as a challenge for Law
Our society is becoming more and more individualistic, but lacks solutions to challenges that are collective or even universal: globalisation, climate change, digitalisation. The digital transformation leads to the datafication of societies and is a collective phenomenon with many open questions and not nearly enough sufficient answers. In my lecture, I will explain why data-driven digital technologies are caused by and emerge from collective structures, simultaneously unfold collective dimensions of impact and why this is a challenge for law.
Mittwoch, 8. Februar 2023, 18.00 Uhr
Carina Prunkl, University of Oxford, Institute for Ethics in AI
Can we, will we, and should we have AI judges?
DER TERMIN MUSS LEIDER ENTFALLEN!
„ConTrust“ – ein Clusterprojekt des Landes Hessen am Forschungsverbund „Normative Ordnungen“ der Goethe-Universität Frankfurt am Main, Frankfurter Gespräche zum Informationsrecht des Lehrstuhls für Öffentliches Recht, Umweltrecht, Informationsrecht und Verwaltungswissenschaften der Goethe-Universität Frankfurt am Main