CHSPR Seminar | When AI Sounds Right but Gets Research Wrong: Demonstrations from Health Research Workflows

Speaker: M. Ehsan Karim, UBC School of Population and Public Health

AI tools are increasingly acting like research assistants: finding papers, summarizing evidence, writing code, and drafting/reviewing manuscripts. Their outputs can be fast, polished, and persuasive — but sometimes the most convincing answer is the one that most needs checking. This seminar uses concrete demonstrations from statistical and epidemiologic research workflows to show why LLM audit literacy is becoming an essential research skill. Rather than offering a technical training session, the talk highlights why verification, documentation, and domain expertise matter before AI-generated outputs are used in research and decisions.

M. Ehsan Karim’s research focuses on biostatistics, advanced epidemiological methods, and causal inference, including the evaluation of high-dimensional propensity score and machine learning applications. His recent work addresses the intersection of artificial intelligence and health research.