Large Language Models for Bio-image Analysis

Large Language Models (LLMs) change the way how we do science, especially how we write scientific data analysis code. This talk starts with an introduction into the topic and into benchmarking LLMs for bio-image analysis code generation. I also explain two projects giving access to LLMs in a user-friendly manner: bia-bob and git-bob. Such tools facilitate the use of LLMs in scientific data analysis and code generation. I furthermore outline why competing with commercial LLM service providers in developing better models and systems is challenging and how we can profit from the developments these service providers achieve.