Large Language Models: Opportunities, Challenges and Pitfalls

Abstract

Ever since ChatGPT was released, Large language models (LLMs) have captured the imagination of researchers and users worldwide and led to a capacity race. We examine their utility for detecting and explaining deceptive attacks such as fake news, fake product reviews, and phishing. Then, we consider the challenges in the context of the detectability of LLM-generated content and in solving orthographic constraints. We conclude with our rexperiences with LLMs at the University of Houston and at the Naval Surface Warfare Center (NSWC), Dahlgren. Directions for future research and insights into LLMs revealed by our research will be discussed including the consequences of aligning LLM models for different purposes, such as 1) ability to mimic humans on natural language tasks and 2) safety.
(This talk is based on joint work with Dainis Boumber, Nachum Dershowitz, David Marchette and Bryan Tuck)

Speaker

Rakesh Verma is a Professor of Computer Science at the University of Houston (UH), where he pioneered a Security Analytics course over a decade ago. He received UH's 2013 Lifetime Mentoring Award for his extensive mentoring of undergraduate students in research. He is the author/co-author of 150+ peer-reviewed publications in leading venues and the book "Cybersecurity Analytics" (CRC Press, 2019). His latest book is "Life: A User's Manual" (Amazon, 2026). He is a cybersecurity section associate editor of the Frontiers in Big Data journal; and a co-organizer of the ACM Annual International Workshop on Security and Privacy Analytics (IWSPA) since 2015. His research interests include formal methods (especially equational logic programming), algorithm design and analysis, AI, and computer security. His research has been funded by the National Science Foundation, the Fulbright Foundation, the Department of Defense, the US Department of Transportation, the Office of Naval Research, and several companies.  He received a B. Tech (Gold Medal, highest honors) in Electronics Engineering from IIT Varanasi and MS/PhD in Computer Science from SUNY Stony Brook, USA, where he was a Catacosinos Fellow.