GPT-5 Pro Unlocks Three-Year Immunological Mystery, Revolutionizing T-Cell Research

Immunologist Derya Unutmaz, a professor at The Jackson Laboratory and the University of Connecticut, has leveraged OpenAI's GPT-5 Pro to resolve a complex three-year-old mystery concerning the precise impact of glucose on T cell development and specialization. This breakthrough, made possible by the AI model released in late 2025, underscores the transformative potential of advanced large language models in accelerating fundamental biological discovery.

The core of the puzzle stemmed from an experiment conducted in 2022, which left researchers without a complete understanding of how certain glucose manipulations affected T cell pathways. GPT-5 Pro provided a critical insight: it suggested that deoxyglucose interfered with the construction of the IL-2 protein. This interference, the model deduced, was crucial in preventing T cells from differentiating into inflammatory-response Th17 cells, a key finding with implications for understanding autoimmune diseases and immune responses.

Predictive Capabilities Validated

Beyond solving existing enigmas, GPT-5 Pro demonstrated remarkable predictive capabilities. Unutmaz utilized the AI to simulate an entirely new experiment, predicting the outcome of CD8+ T cells' efficacy in targeting lymphoma cells. The model’s prediction proved accurate, validating its capacity not just for retrospective analysis but also for guiding prospective experimental design and forecasting biological responses. This ability to accurately simulate and predict complex cellular interactions marks a significant leap for AI in scientific methodology.

Unutmaz's engagement with advanced AI tools is not new; he has previously experimented with models like Codex and GPT-5.2 Deep Research. These earlier applications focused on compiling extensive cancer mutation datasets and generating preliminary research materials. The successful application of GPT-5 Pro to a long-standing immunological question highlights a significant maturation in AI's utility, moving from data aggregation and content generation to direct problem-solving and experimental prediction in highly specialized scientific domains. This development signals a new era for AI-driven research, where sophisticated models act as integral collaborators in unraveling complex biological processes.