Responsible Data Science (Sommersemester 2024)

Lecturer: Dr. Simon David Hirsbrunner

General Information

• Weekly Hours: 2
• Credits: 3
• Graded: yes
• Enrolment Deadline: 01.04.2024 - 15.04.2024
• Examination time §9 (4) BAMA-O: 24.04.2024
• Teaching Form: Seminar
• Enrolment Type: Compulsory Elective Module
• Course Language: English
• Maximum number of participants: 20

Programs, Module Groups & Modules

Description

The seminar provides students with strategies, knowledge and tools to reflect and address societal and ethical risks in data-intensive projects (data engineering, machine learning, etc.).
Please check the Github-website of the seminar for up-to-date information and links toresources: https://github.com/simonsimson/responsible-data-science.

On the one hand, various ethical challenges are introduced (e.g. discriminatory bias and fairness, opacity and transparency, reliability and security, contestability and accountability) and examples discussed by means of practical use cases and interdisciplinary literature. On the other hand, students will learn about approaches and tools that can be used to address corresponding problems in the context of project development (value-sensitive design, stakeholder mappings, risk analysis, value scenarios, simulations). The seminar invites applicants from multiple research fields to facilitate a rich discussion of diverse perspectives and different value concepts within in the seminar.

Thematically, the seminar focuses on the application of AI in high risks scenarios (e.g. facial recognition, intelligence and evidence gathering in law enforcement, AI-supported personnel recruitment, autonomous weapon systems, AI in healthcare) and on ethical aspects related to Generative AI (opacity of foundation models, disinformation, deepfakes, cultural bias, environmental impact).

Requirements

Affinity to interdisciplinary perspectives (computer science, data science, human-computer interaction, social sciences, ethics, law), willingness to read scientific texts from different disciplines, ideally existing experience in the implementation of technology projects.

Literature

Literature from applied ethics research, social sciences (critical algorithm and data studies), computer science (responsible AI, human-centered computing), regulation and standards (EU AI Act, ISO, VDE Spec) and civil society perspectives (NGO statements, journalistic media).

Learning

Weekly seminar each with input lectures by the instructor as well as by guests; short presentations by students based on the literature provided; joint exercises, discussions and simulations based on practical examples; development of analyses of ethical risks and addressing strategies for different scenarios of technical systems development and deployment. Each session of the seminar will focus on one specific ethical issue area and on concrete strategies and tools for addressing corresponding risks.

Examination

The grade is based on a group presentation (2 students) of 20 - 25 min. incl. handout (30% of the grade) and a paper of 6 pages per person (70% of the grade). The group presentations will each address one specific literature piece and related ethical issues. For the paper, students address one ethical dilemma in an application area of AI or data engineering using the concepts and methods learned in the seminar. The lecturer will provide suitable subject areas from which the students may choose from.

Dates

10.04.2024, 13.30 – 15.00 h
Session 1: Introductory Session with Assignment of Presentation Topics

24.04.2024, 13.30 – 16.45 h
Session 2: Applied Ethics and Value-Sensitive Design

15.05.2024, 13.30 – 16.45 h
Session 3: Discrimination, Fairness and Diversity

22.05.2024, 13.30 – 16.45 h
Session 4: Privacy and Informational Self-Determination

29.05.2024, 13.30 – 16.45 h
Session 5: Deep Fakes and Disinformation

05.06.2024, 13.30 – 16.45 h
Session 6: Human Oversight and Contestability

12.06.2024, 13.30 – 16.45 h
Session 7: Transparency, Documentation and Accountability

26.06.2024, 13.30 – 16.45 h
Session 8: Paper Review Session and Wrap Up

Room: D-E.9/10

Contact: Simon.Hirsbrunner(at)guest.hpi.de

Zurück