Lecture: Creating Equitable Value with AI
Join us for a session in the company of three exciting researchers sharing their work and perspectives on bias and fairness in AI – and the dilemmas in ensuring responsible and value-creating AI.
Read more
Fairness in artificial intelligence is crucial to ensure that AI-systems serve all individuals equitably, preventing discrimination and promoting trust. Bias in AI-systems can compromise fairness and value creation in multiple ways, e.g., by perpetuating historical injustices in the training data. Both fairness and bias are nuanced concepts, which require careful consideration of the context in which AI-systems are developed and deployed. Join us for a session in the company of three exciting researchers sharing their work and perspectives on bias and fairness in AI – and the dilemmas in ensuring responsible and value-creating AI.
At first, Professor Aasa Feragen will share her research on AI-systems in healthcare. This involves specific examples of the challenges regarding fairness and bias, but also how these can be adressed.
Secondly, Professor Ira Assent will use recommender systems as an example to illustrate fairness issues. Recommender systems influence how we find information and connect with others, but they can unintentionally favor certain groups, leading to unfair outcomes. As Ira and colleagues show in a recent study of such recommender systems for politically polarized data from Twitter, political alignment affects the relevance and variety of recommendations. Finally, Ira will explore other current fairness aspects in AI.
In the last talk of the session, Professor Thomas Moeslund will discuss some of the concepts in current and upcoming regulations of the use of AI. These include notions like ‘Human-centric AI’, ‘Fair AI’, ‘Robust AI’ etc. While such concepts seem reasonable it becomes a bit more fuzzy when defining how to make these auditable. Thomas will dwell into these matters and exemplify the current gap between high-level notions and practical solutions.
Important Dates
Facts
Programme
