A Human-centric Approach to Researching Social Bias in NLP Technologies

Date: 6 November 2023

Time: 3:10pm-4:15pm

Venue: Informatics Forum 1.15

Authors/Speakers: Eddie Ungless

Abstract: Much current research into social bias in Natural Language Processing – that is, the tendency for NLP technologies to reflect human biases such as sexism – suffers from a superficial understanding of the problem. The issue of bias is treated as a mathematical kink that needs to be ironed out, after which the harm the model does will be irrefutably reduced. I argue this superficial understanding of bias will lead us to superficial solutions which ignore the role of human behaviour in determining the harm done by technology. In this talk, I advocate for a human-centric approach to studying bias in NLP, one which focuses on human choices, human identities and human behaviour, to give a more complete understanding of the true impact of AI. I present two papers as case studies for the benefits of this approach, each addressing queerphobia in publicly available NLP tools. 

About the Speaker: Eddie L. Ungless is a final year PhD student in the Centre for Doctoral Training (CDT) in NLP , funded by the UKRI. He has an interdisciplinary background spanning linguistics, psychology, digital media strategy and computer science. His work addresses social bias in NLP technologies, wherein he champions an approach that builds on social science research to centre human experiences in our understanding of AI harms. You can find out more about his interests, along with links to his published work, on his blog: https://mxeddie.github.io/