Dia: 03/07/2024
Hora: de 12.00h a 13.00h
Lloc: sala 52.701, Campus del Poblenou UPF, edifici Roc Boronat
 

Abstract:
Language and morality are intimately related. Morals can guide language use and development, while language can reveal people’s morals and moral perception. Advances in artificial intelligence (AI) have offered a new opportunity for studying these relations through the lens of computation. In this talk, I describe recent work that focuses on exploring the interplay of language, morality, and computation. First, I show that computational methodologies drawing on diachronic word embeddings support automated inference of historical shifts in moral sentiment toward concepts such as slavery over centuries and the cognitive processes underlying the evolution of moral lexicon. Next, I show that morally relevant linguistic inputs serve important functions to probing moral emergence and bias in AI systems such as large language models. I close by discussing ongoing and future work on the synergy of language, morality, and computational intelligence.