Course Content
The seminar will cover the different aspects of AI sustainability. Students will get to engage with a choice of topics spanning a variety of modern perspectives and assessments of current state-of-the-art approaches. Respectively engaged literature serves as the basis to practice presentations and hold critical discourse among peers. Example questions for the specific covered dimensions of AI sustainability are provided below. At the start of the semester, an overview presentation with respect to the different dimensions will be given and students will get to pick their preferred topics from a curated list of recent papers.
Learning Outcome
AI is conjectured to hold transformative economic potential and benefit society in several ways. Among countless aspects, it promises to aid in the climate crisis, foster food stability, democratize knowledge, and empower humans with unprecedented capabilities. Throughout the seminar, students will learn to critically engage with current perspectives on AI’s impact, analyze the validity of claims with respect to economical, environmental and social sustainability, and separate opportunities from unaddressed harms. As such, students will improve their ability to engage with multi-faceted modern literature and participate in broader AI discourse.
Example questions and dimensions of sustainability
Computational and environmental sustainability: how efficient are AI algorithms and how flexible are current workflows? Where does compute excel and where are the present limits? What are AI’s contributions to the environment and in which ways are adverse effects entailed?
Economic sustainability: What is the potential of AI to transform industries and contribute to financial growth? Which current AI use-cases are profitable and how efficiently are AI assets used in practice? What are applications where AI fell short and how can we spot hype?
Social sustainability: How inclusive are AI approaches and how do they represent diverse cultures? What outcomes do AI systems hold for the average population and marginalized community? How can AI interface with moral systems and law?
Human sustainability: What role does AI play in education and democratization of knowledge? What are respective risks of current AI systems? In which ways is AI able to support humans instead of replacing them?
Seminar Schedule
08.05, 10:15: Consent in Crisis: The Rapid Decline of the AI Data Commons (NeurIPS 24)
15.05, 10:15: Unmasking AI (book)
15.05, 11:00: A theory of appropriateness with applications to generative artificial intelligence (arXiv 24)
22.05, 10:15: Building machines that learn and think with people (Nature Human Behavior 24)
22.05, 11:00: Epistemic Injustice in Generative AI (AIES 24)
05.06, 10:15: More than a Glitch (book)
05.06, 11:00: Opportunities, challenges, and benefits of AI innovation in government services: a review (Disc. Artif.. Intell. 24)
12.06, 10:15: Biological underpinnings for lifelong learning machines (Nature Machine Intelligence 22)
19.06, 10:15: The Impact of Artificial Intelligence on Productivity, Distribution, and Growth (OECD Artificial Intelligence Papers 24)
26.06, 10:15: How much energy will AI really consume? The good, the bad and the unknown (Nature News Feature 25)
03.07, 10:15: The PRISM Alignment Dataset: What Participatory, Representative and Individualizsed Human Feedback Reveals About the Subjective and Multicultural Alignment of LLMs (NeurIPS 24)
03.07, 11:00: Medical Hallucinations in Foundation Models and Their Impact on Healthcare (arXiv 25)
10.07, 10:15: (Unfair) Norms in Fairness Research: A Meta-Analysis (arXiv 24)
10.07, 11:00: Tackling Climate Change with Machine Learning (ACM Computing Surveys 22)