University of Cambridge > Talks.cam > Language Technology Lab Seminars > Assessing the Social Capacity of Large Language Models

  Browse A-Z

Log in

University Account External (via Google)

🔐 Log In

Information on

Subscribing to talksdetails Finding a talkdetails Adding a talkdetails Disseminating talksdetails Help and Documentationdetails

ℹī¸ Information

Assessing the Social Capacity of Large Language Models

Download to your calendar using vCal

If you have a question about this talk, please contact Panagiotis Fytas .

Much of language is social: nearly all of what we communicate is directed to other people in a particular social context. Yet, our traditional models for natural language understanding have typically operated independently of this social context. With large language models (LLMs) now being used in more interpersonal contexts, we ask to what degree do these models understand the social world in which they operate. In this talk, I will describe three studies aiming to measure the social capacity of LLMs from different angles: their ability to recognize different types of social information in messages, their ability to use the social context to correctly interpret a context-sensitive message, and their potential for innate personality-like characteristics. Through these examples, I will highlight the importance of social information for improving natural language processing models.

This talk is part of the Language Technology Lab Seminars series.

This talk is included in these lists:

Note that ex-directory lists are not shown.

Other lists

Cambridge University Engineering Department, CBL, BE-438 (http://learning.eng.cam.ac.uk/Public/Directions) Average on Sundays: 7286.88392857 kWh Bateman Auditorium, Gonville & Caius

Other talks

Bacterial Physiology During Growth Arrest: Can It Be Targeted by Anti-Bacterial Therapeutics? Organiser Welcome Content, Caching and Kubernetes: Performance and User Experience in the Browser Title to be confirmed Group Work Discussion II continued
 

Š 2006-2025 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity