AI is moving from chat assistant to research partner. I will discuss large language models and their current limitations, including uneven capabilities, hallucinations, the need for guardrails, and the risk of epistemic debt. I will share lessons from deploying AI across Garvan and present case studies in bioinformatics and computational biology, from agentic coding and custom MCP connectors to skills-based analysis workflows and discuss the emerging role of AI as a co-investigator.
Associate Professor Leonard Goldstein is a computational biologist at the Garvan Institute of Medical Research, where he heads the Data Science Platform and leads the institute’s AI strategy. He is a Conjoint Associate Professor in the School of Clinical Medicine at UNSW Sydney. Trained in mathematics and computational biology at the University of Cambridge, he spent eight years at Genentech in South San Francisco developing bioinformatics methods for transcriptome and single-cell analysis and leading biomarker analysis of oncology clinical trials. His research spans cancer genomics, single-cell and spatial biology, and immunology, with a current focus on AI-enabled scientific discovery.