The objectives and expected outcomes of this seminar are as follows:
- Participants will gain a deeper understanding of the challenges associated with AI alignment, including biases, interpretability, and value misalignment.
- The seminar will facilitate interactive discussions to develop concrete solutions and strategies for addressing AI alignment challenges.
- Participants will have the opportunity to network with leading experts and build connections that can lead to ongoing collaborations and partnerships
This event is a collaboration between Universitetsfondet and UiB AI.
Read more and register by 26.05 on UiB AI's website.
Programme:
09:00-09:30: Registration and coffee
09:30-09:40: Introduction by moderator Samia Touileb, Department of Information Science and Media Studies at University of Bergen
09:40-10:10: Tom Potter, BBC: "Building Responsible AI Tools for Public Service Journalism"
10:10-10:40: Leonora Onarheim Bergsjø, HiØ og UiA: "Ethical Risk Assessment of AI"
10:40-11:00: Coffee break
11:00-11:30: Rune Nyrup, Aarhus university: "Rigorous Evaluation Methods for AI Explainability: How can philosophy and computer science collaborate?"
11:30-12:00: Holli Sargeant, St. John's College, Cambridge: "Encoding Equality: The Incompatibility of Algorithmic Logic and Law"
12:00-12:45: Lunch (included)
12:45-13:15: Jan Broersen, Utrecht university: "Do LLMs mean what they say?"
13:15-13:45: Davide Liga, University of Luxembourg: "Mechanistic Interpretability and Moral Stance in Large Models"
13:45-14:00: Coffee break
14:00-14:30: David Samuel, University of Oslo: " Aligning Norwegian LLMs to user preferences and expectations"
14:30-15:00: Ryan Marinelli, University of Oslo: Something about safety or security
15:00-15:45: Wrap up and discussions