Toward Responsible Natural Language Processing : Ideal, Illusion, or Imperative?

Large language models have profoundly transformed natural language processing and are increasingly reshaping work, knowledge production, and social organization, yet they remain misaligned with societal values and demand substantial computational resources. In this seminar, I will present my research on responsible NLP, structured around two central pillars: fairness and frugality. Through a scientific overview of selected recent and ongoing works, I will discuss methods to assess and mitigate alignment failures, and to develop resource-efficient approaches that promote more sustainable NLP systems.