Arne Vanhoyweghen is an AI researcher and data scientist working to make large language models more reliable, interpretable, and deployable in high-stakes settings. He is a postdoctoral researcher at the Data Analytics Lab of Vrije Universiteit Brussel and a guest professor in Business Technology and Operations at the Faculty of Social Sciences & Solvay Business School.

His research investigates how LLMs reason, where they fail, and how internal signals in their outputs can be harnessed to detect uncertainty and improve calibration. He applies these insights to design structured AI workflows for education, encompassing automated assessment, feedback generation, and scalable learning assistance. These systems integrate advanced prompting strategies, multi-pass evaluation, and human-in-the-loop verification to ensure reliability, fairness, and pedagogical alignment.

His expertise spans LLM evaluation, uncertainty detection, chain-of-thought analysis, AI-assisted assessment, and human–AI collaboration systems. In parallel, he conducts research in Ergodicity Economics, where he studies decision-making under non-ergodic dynamics and long-run growth processes, linking formal economic theory with computational modeling.