CLARIN 2025 Keynote - Lonneke van der Plas - Creativity and AI

Описание: ➡️Event details, proceedings and materials:
https://www.clarin.eu/event/2025/clar...
➡️Download presentation slides:
https://www.clarin.eu/media/9163

➡️Follow us on:
🔗https://www.clarin.eu/
🔗  / clarin-eric  
🔗https://github.com/clarin-eric

--------------------------------------------------
Lonneke van der Plas is associate professor in NLP at USI Università della Svizzera italiana in Lugano (CH). She leads the Computation, Cognition & Language (CCL) group which she launched at the Idiap Research Institute in Martigny, in 2021. She was associate professor at the University of Malta from 2014-2020, and before that, she was junior professor at the University of Stuttgart. She earned her PhD from the University of Groningen. Research in the CCL group is highly interdisciplinary, including collaborations with cognitive scientists, linguists, the life sciences, finance and business. Lonneke has over 70 publications on the following topics: computational creativity, multi-lingual natural language processing (including low-resource languages), diachronic modelling of language, (medical) question answering, NLP for HR, semantic role labelling etc.

--------------------------------------------------
ABSTRACT:

People adapt to the world that is constantly evolving. They find relationships between pairs of seemingly very distantly related concepts and use these for creative problem solving or to come up with new ideas. Also, they produce an immense variety of creative output. Creativity is an important ability for diversification and survival. These aspects of creativity have been under-explored in NLP research and AI in general. Some forms of creativity in humans can be approximated with computational models of language. Moreover, creative tasks require a high level of abstraction that allows the system to cut across the more straightforward patterns found and are therefore a perfect testbed for measuring the generalisation power and, arguably, the actual level of intelligence found in current systems. However, research in creativity faces several challenges. One of these is the difficulty of defining creativity and partly related to that, the issues in evaluating creative behaviour and products. In this talk, I will first motivate the need to investigate computational creativity and briefly discuss the challenges. After that, I will showcase two pieces of recent work, one attempts to bridge cognitive science and NLP for modelling analogical reasoning and the other assesses the creativity in the output of large language models and proposes a method to improve this aspect.

--------------------------------------------------
Event date: [02/10/2025]

🗓️Our events calendar:
🔗https://www.clarin.eu/events

---------------------------
🕑Timestamps:
00:00 – Creativity and AI