Research Talk Announcement

Department of Information Sciences and Technology

Via Zoom

https://gmu.zoom.us/j/98665607139?pwd=ERDUuLPQvlPfD57lx8Y94s2P0xLAWU.1

Meeting ID: 986 6560 7139

Passcode: 418223





Title: Novel Frameworks for Quantifying Political Polarization and Bias Auditing Large Language Models

Speaker: Dr. KhudaBukhsh

Date and Time:  June 4, 2024, at 11:00 AM

Meeting Room: Virtual Only



Abstract:

This talk is divided into two parts. Each part summarizes a broad line of natural language processing (NLP) research outlining a new framework. In the first part, I will describe a new methodology that offers a fresh perspective on interpreting and understanding political and ideological biases through machine translation. Focusing on a year that saw a raging pandemic, sustained worldwide protests demanding racial justice, an election of global consequence, and a far-from-peaceful transfer of power, I will show how our methods can shed light on the deepening political divide in the US. The second part of the talk presents a novel toxicity rabbit hole framework to bias audit large language models (LLMs). Starting with a stereotype, the framework instructs the LLM to generate more toxic content than the stereotype. Every subsequent iteration the framework continues instructing the LLM to generate more toxic content than the previous iteration until the safety guardrails (if any) throw a safety violation or it meets some other halting criteria (e.g., identical generation or rabbit hole depth threshold). Our experiments reveal highly disturbing content, including but not limited to antisemitic, misogynistic, racist, Islamophobic, and homophobic generated content, perhaps shedding light on the underbelly of LLM training data, prompting deeper questions about AI equity and alignment.



Speaker's Bio:

[cid:image001.jpg@01DAB596.B17A11E0]Ashique KhudaBukhsh is an assistant professor at the Golisano College of Computing and Information Sciences, Rochester Institute of Technology (RIT). His current research lies at the intersection of natural language processing and AI for Social Impact as applied to: (i) polarization in the context of the current US political crisis; (ii) globally important events arising in linguistically diverse regions requiring methods to tackle practical challenges involving multilingual, noisy, social media texts; and iii) auditing AI systems and platforms for unintended harms. In addition to having his research been accepted at top artificial intelligence conferences and journals, his work has also received widespread international media attention that includes coverage from The New York Times, BBC, Wired, Times of India, The Daily Mail, VentureBeat, and Digital Trends.

Thank you!

Sincerely,

Sarah Alharshan

Administrative Specialist
Department of Information Sciences & Technology
College of Engineering and Computing
Phone: (703) 993-4871
E-mail: [log in to unmask]<mailto:[log in to unmask]>