Large language models (LLMs) are artificial intelligence systems that serve as the foundation for various conversational agents like ChatGPT. They generate responses that can be morally evaluated.
This two-day exploratory workshop aims to collectively test the “moral psychology” of these large language models that come across as black boxes. Do the responses exhibit specific moral biases? Are there “styles” that differentiate LLMs from human beings? What obstacles hinder the moral evaluation of LLMs?
The workshop will begin with a presentation outlining the current state of research on the subject. It will then involve various exercises: prompt creation, response evaluation, comparisons, and “flaw” identification. Participants will be actively involved in hands-on activities, contributing to imaginative solutions for a better understanding of the ethics of LLMs.
(By invitation only)
