La Mobilière Lecture Series 2025: Prof. Dr. Thomas Hills on Algorithmic Psychotherapy: A Humanistic Approach to Understanding Biases in AI

We are pleased to announce the upcoming La Mobilière Lecture Series 2025, where we will welcome Prof. Dr. Thomas Hills from Universität of Warwick, as our distinguished guest speaker. Prof. Dr. Thomas Hills will present her research on "Algorithmic Psychotherapy: A Humanistic Approach to Understanding Biases in AI"
Join us for an engaging discussion on this pressing topic and the opportunity to exchange ideas with fellow researchers and professionals. We look forward to seeing you there!



Abstract:
All knowledge is biased. Large Language Models (chatbots), like other deep neural networks, embed their biases in numerical representations that cannot be observed. One cannot look at the code to determine what they do and do not know. To understand them, we must anticipate bias, in the same way we might anticipate hidden motivations and beliefs in humans, which are sometimes even hidden to the humans themselves.
In his talk, Thomas Hills will discuss various algorithmic biases and describe approaches to understanding these biases using prompts inspired by humanistic psychology. He will present one experimental case study in which the results of humans using this approach outperform humans with or without chatbot assistance.