Assessing AI’s cognitive abilities for scientific discovery in the field of systems vaccinology

The advent of large language models (LLMs) has transformed academic research by accelerating hypothesis generation and data analysis. LLMs can help researchers uncover patterns and insights from vast datasets to foster innovative scientific discovery. However, questions arise regarding the creative capacity of artificial intelligence (AI), especially in biologically complex fields such as vaccinology. This study evaluates the ability of LLMs to generate hypotheses, design experiments, and infer broader biological principles through a proposed framework called “The Creation Game.” Using three case studies—general control nonderepressible 2 (GCN2)’s role in dendritic cell antigen presentation via stress response, sterol regulatory element–binding protein (SREBP)’s influence on metabolic responses, and Toll-like receptor 5 (TLR5)’s connection to microbiota-driven vaccine efficacy—we assessed AI’s accuracy, logic, and creativity. The findings underscore the potential of LLMs to accelerate vaccine research while emphasizing the importance of ethical oversight. By complementing human creativity, AI could potentially transform hypothesis-driven science, paving the way for tailored vaccination strategies and deeper insights into human immunity.