The Challenge of Detecting AI-Generated Text

Abstract:

Large language models (LLMs) have advanced to a point that even humans have difficulty discerning whether a text was generated by another human, or by a computer. However, knowing whether a text was produced by human or artificial intelligence (AI) is important to determining its trustworthiness, and has applications in many domains including detecting fraud and academic dishonesty, as well as combating the spread of misinformation and political propaganda. The task of AI-generated text (AIGT) detection is therefore both very challenging, and highly critical. In this talk, I will briefly summarize state-of-the art approaches to AIGT detection, including watermarking, statistical and stylistic analysis, and machine learning classification. I will then discuss our research experiments exploring the effect that fine-tuning LLMs has on their subsequent detectability, specifically in the domain of social media.

Bio:

Dr. Kathleen Fraser is an associate professor in the University of Ottawa School of Electrical Engineering and Computer Science. Her research focuses on natural language processing (NLP), artificial intelligence (AI), and the social impacts of these technologies. One aspect of her research examines the social biases and stereotypes that end up encoded in AI systems. She also conducts research around themes of AI Safety, in particular looking at how user modifications to frontier models (such as fine-tuning) impact the safety profiles of those models. She also has a research interest in NLP for healthcare, specifically the automated analysis of speech and language in conditions such as dementia, aphasia, and post-COVID condition (“long COVID”). Dr. Fraser completed her PhD at the University of Toronto in 2016, and her post-doc at the University of Gothenburg in 2018. She worked as a research scientist at the National Research Canada of Canada for 7 years, before moving to the University of Ottawa.

Pricing

• Free

Dates and Times

Contact

Sharif Abuadbba

More information

Register online to receive the login details.