1. Przejdź do Treść
  2. Przejdź do Menu główne

Data publikacji w serwisie: 16 maja 2023 r.

SLE Online Lecture Series: The Linguistics of Deep Learning: ChatGPT and Large Language Models by prof. Willem Zuidema

The Linguistics of Deep Learning: ChatGPT and Large Language Models

Willem Zuidema (University of Amsterdam)

Thursday 25 May

5:30 – 6:30 PM (CET – Paris – Brussels).

The platform is Zoom, and the link to the Zoom-meeting can be accessed by logging in to the members’ area on the SLE website. If you are not SLE member, you can receive the link after registering here.    

The recording of the lecture will be made available afterwards.

Below are the title and the abstract of the lecture:

The Linguistics of Deep Learning: ChatGPT and Large Language Models

Willem Zuidema, Associate Professor in Natural Language Processing, Explainable AI and Cognitive Modelling at the Institute for Logic, Language and Computation of the University of Amsterdam. Website

Abstract

Large Language Models, such as those underlying ChatGPT, have shown the world how far deep learning for natural language processing has advanced. How do these models work, and to what extent have they really discovered the underlying regularities of natural languages? I will briefly discuss the architecture and training regimes of these models, and ways to probe their capabilities. I will argue that the many differences with how humans learn and use language do not mean that LLMs are irrelevant for linguistics and cognitive science; on the contrary: they offer new opportunities, new risks, and new perspectives on old dilemmas.