Jacopo
Amidei

In recent years, I have been working as a PhD researcher and Research Fellow in Mathematical Logic and Computational Linguistics. In November 2012 I took on a PhD in Philosophy and Mathematical Logic at the Scuola Normale Superiore (SNS) of Pisa. My main research areas have been Computability Theory and Probability Theory. In 2016 I won an Erasmus+ Mobility Consortium Talent at Work Traineeship Mobility grant, which was used for a traineeship (from March to September) at the AI software solutions company ExB in Leipzig (Germany). My interest in Machine Learning stems from that experience which strongly motivated me to pursue a career in Machine Learning for NLP. Accordingly, I began a PhD in Computational Linguistics at The Open University (UK). For my PhD, I focused on the Reliability of Human Evaluation of Natural Language Generation (NLG) Systems. After my PhD, I won a position, running from Jan. 2021 to Apr. 2023, as a research fellow on the Opening Up Minds: Engaging Dialogue Generated from Argument Maps project. The project, which addressed the problem of fostering open-mindedness in debates, was a collaboration among The Open University, the University of Sheffield, the University of Cambridge and the Cambridge Research Laboratory, Toshiba Europe. Since June 2023 I have been a Postdoctoral Fellow at Universitat Oberta de Catalunya (UOC) where I joined the Artificial Intelligence and Data for Society (AID4So) group. In this new appointment, I focus my research on the ethical application of AI. In this vein, I started an ongoing collaboration with psychologist and researcher Rubén Nieto (full-time professor at UOC). This collaboration focuses on the intersection of AI (particularly LLMs), and psychology. To date, our focus has mainly been on: 1) assessing classical psychological constructs in OpenAIs GPT models in a multilingual setting, 2) simulating personas and evaluating the psychometric properties of classical questionnaires designed to assess personality, also in this case in a multilingual setting and 3) use LLMs to aiding clinicians in their assessment of patients with pain.