Generative AI tools like ChatGPT are spreading rapidly into professional work, yet little is known about how managers use them for their decisions: choices marked by ambiguity, competitive pressure, and accountability. Experimental research shows productivity gains for routine knowledge work (Noy & Zhang, 2023; Brynjolfsson et al., 2025) but also reveals a sharp "jagged frontier": AI can reduce performance on tasks that fall outside its competence, often without the user noticing (Dell'Acqua et al., 2023).

Strategic decisions are the domain where this frontier is hardest to locate and where the costs of misjudgment are highest. Recent work suggests that consulting LLMs may push managers toward overly rigid problem framings (Galli Geleilate & Humberd, 2024), and that AI may fundamentally reshape the cognitive processes of search, representation, and aggregation in decision making (Csaszar et al., 2024).

The thesis is supposed to answer the research question: "How do managers perceive the usefulness and risks of LLMs for individual decision-making, and how do these perceptions shape actual patterns of use?"

If you are interested in this thesis, feel free to contact Tobias Kotzian (tobias.kotzian@unisg.ch).