25 November 2026
Can We Trust AI If Humans Are in the Loop?
Prof. Dr. Frauke Kreuter
Ludwig-Maximilians-Universität München, DE
Whose judgments shape AI — and how AI shapes ours
AI systems are trained with human feedback, and people are increasingly asked to rely on AI-generated answers. This makes the idea of keeping a “human in the loop” sound reassuring. But it leaves out two important questions: Who are the people whose judgments shape the model in the first place? And what happens when other people are later asked to check what the model produces?
In this lecture, Frauke Kreuter looks at both sides of this relationship. Experiments on AI-assisted data work show that human oversight is not automatically a safeguard. People are quite good at spotting some machine errors, but much less likely to catch others, particularly when an incorrect answer looks plausible.
There is also an earlier, less visible human contribution. The feedback used to train and adjust AI systems often comes from relatively narrow groups of annotators. This raises a basic question familiar from survey research: whose views are represented, and who is missing? Kreuter will discuss how methods from statistics and social science can help make such questions measurable, including how well models reflect differences across groups and changes over time.
Knowing what people think, however, is not the same as deciding what an AI system ought to say or do. The lecture therefore also considers how people and AI might work together differently — not just by having machines produce answers that humans approve, but by allowing people to remain involved as questions, assumptions and analyses develop.
The aim is to get beyond the simple idea that adding a human checker makes an AI system trustworthy.
About the Speaker
Professor Frauke Kreuter holds the Chair of Statistics and Data Science at LMU Munich, Germany and at the University of Maryland, USA. She is Co-Director of the Social Data Science Center (SoDa) and faculty member in the Joint Program in Survey Methodology (JPSM). She is an elected fellow of the American Statistical Association, and received the Warren Mitofsky Innovators Award of the American Association for Public Opinion Research in 2020. In addition to her academic work, Professor Kreuter is the Founder of the International Program for Survey and Data Science (IPSDS), developed in response to the increasing demand from researchers and practitioners for the appropriate methods and right tools to face a changing data environment; Co-Founder of the Coleridge Initiative, whose goal is to accelerate data-driven research and policy around human beings and their interactions for program management, policy development, and scholarly purposes by enabling efficient, effective, and secure access to sensitive data about society and the economy; and Co-Founder of the German language podcast Dig Deep.
photo credit: SoDa, LMU, cropped
Lecture Time and Location
Wednesday, 25 November 2026
18:15–19:45 (CET)
Flügelbau Ost, 2. OG, Raum O 221
Edmund-Siemers-Allee 1
20146 Hamburg
If you like to join us virtually, please register with your email-address and name.
TAMING THE MACHINES
Register for the webinar, explore this semester’s program, browse past lecture programs, and access recordings of previous talks. Stay informed, and catch up anytime!
Photo: Alina Constantin / Better Images of AI / Handmade A.I / CC-BY 4.0