Cancer tested by generative AI

Cancer put to the test by generative AI

March 16, 2026

By Hugo Jalinière THE Subscribers

New algorithms capable of interacting with complex biological data are beginning to transform oncology research and are outlining a new way of understanding and treating the disease.

Breast cancer

Colorized scanning electron micrograph (SEM) of a breast cancer cell.

Photo by STEVE GSCHMEISSNER/SCIENCE PHOTO / SGS / SCIENCE PHOTO LIBRARY VIA AFP

This article is taken from the monthly magazine Sciences et Avenir n°949, dated March 2026.

What if we could query a tumor like we query a search engine? The idea may seem far-fetched. Yet, this is precisely the type of artificial intelligence model presented last October by an American team from Yale University and Google's DeepMind lab (creators of the Alphafold program, the subject of the 2024 Nobel Prize in Chemistry). Its name: Cell2Sentence-Scale 27B (or C2S-Scale), which can be translated as "from cell to sentence – scale of 27 billion" – representing the number of parameters taken into account by the model.

Generative AI Artificial intelligence in health

en_USEnglish