Master’s Thesis: Human-AI Collaboration in a Realistic Decision-Making Task

10.08.2026, Studentische Hilfskräfte, Praktikantenstellen, Studienarbeiten

Motivation

Terms like human-AI collaboration, Human-in-the-Loop, and human oversight have become common buzzwords in discussions about how AI can be deployed responsibly. However, simply pairing humans with AI is not sufficient - we have to understand how people work with AI and how to best design AI for human use, e.g., how AI affects human decision-making. For instance, many complex decisions, like choosing between apartments, evaluating job applicants, or selecting medical treatments, are difficult because they involve multiple criteria and no single correct answer. Instead, decision-makers must weigh competing and sometimes conflicting factors. While AI-assisted decision-making is a highly active field of research, most studies rely either on overly simple decision-making tasks, which limits the real-world applicability of results, or study complex real-world settings, which are hard to control and manipulate.

Your Task

In this thesis, you will design and implement a study task environment that is rich enough to mirror the complexities of real-world decision-making, while simultaneously being accessible to a broad group of participants and controllable enough for experimental manipulations. You will further design and implement multiple LLM-based decision support tools and conduct user studies within your task environment to understand how people work with different kinds of decision support under realistic task conditions.

Your Profile

• Strong programming skills, ideally web development.
• Strong interest in human factors of AI.
• Experience in conducting user studies and designing experiments is a plus.
• Experience in UI/UX design is a plus.

Kontakt: zelun.tony.zhang@tum.de