How GPT-5 helped immunologist Derya Unutmaz solve a 3-year-old mystery
TL;DR
OpenAI says GPT-5 Pro helped immunologist Derya Unutmaz revisit a T-cell puzzle that had been sitting unresolved since 2022. The core question: why did T cells exposed to deoxyglucose produce far more inflammatory Th17 cells than T cells grown in a low-glucose environment? GPT-5 Pro proposed that deoxyglucose interfered with production of IL-2, a protein that can normally block T cells from becoming Th17 cells.
Nauti's Take
The interesting part is not that GPT-5 Pro produced a magical answer. It is that an experienced scientist used the model as a mechanism sparring partner on a real, old dataset.
That is where AI starts to matter: connecting adjacent knowledge that sits just outside the researcher’s immediate focus. The catch is sharp: without domain expertise, the same output is just plausible text, not science.
Briefingshow
The case shows where AI can become useful in science: not as a replacement for lab work, but as a hypothesis engine across literature, mechanisms, and experimental traces. In immunology, cancer, and autoimmune research, that can compress weeks or months of work. Still, this is an OpenAI story with obvious product framing, so independent scientific validation matters.
Sources