Prof. Dr. Kosuke Imai holds Distinguished Lecture in spring 2025

The lecture series “MCDS Distinguished Lecture”, introduced in the Spring Semester 2024, continued on March 4 with a lecture by Prof. Kosuke Imai from Harvard University.

As part of the MCDS Distinguished Lecture series, Prof. Imai delivered a talk entitled “Does AI Help Humans Make Better Decisions? A Statistical Evaluation Framework for Experimental and Observational Studies” at the University of Mannheim. Due to the high level of interest, the event was moved to the Wilhelm-Müller-Hörsaal (Room O 131), where around 150 students and researchers attended. The lecture was initiated and moderated by Prof. Marc Ratkovic.

In his talk, Prof. Imai addressed the growing role of artificial intelligence and data-driven algorithms in decision-making processes. While AI systems are increasingly used across many domains, humans often remain responsible for final decisions—especially in high-stakes settings. This raises the fundamental question of whether AI actually helps humans make better decisions.

Prof. Imai introduced a new statistical framework that allows researchers to empirically compare the performance of three decision-making systems: human-alone, human-with-AI, and AI-alone. The approach relies on minimal assumptions and standard classification metrics and is applicable to both experimental and observational studies.

In the second part of the lecture, Prof. Imai demonstrated the framework using a randomized controlled trial on pretrial risk assessment. The results showed that AI-generated recommendations did not improve judges’ decision accuracy in imposing cashe bail. Moreover, replacing human judges with algorithmic systems—such as risk assessment scores or large language models—tended to result in worse classification performance.

MCDS is very pleased with the strong interest in this lecture, sincerely thanks Prof. Imai for his visit, and looks forward to upcoming events in the MCDS Distinguished Lecture series.