The existing pages in this section describe how large language models work internally and how to use them for synthesis and literature work. This page is about something different: AI assistants embedded in the analysis workflow itself — writing pipeline code, converting data, navigating a cluster, drafting documentation.
The interesting questions here are not about capability but about discipline. If an assistant wrote the preprocessing script, what is recorded about how it was produced? If a model suggested an analytic choice, does that belong in the decision log? Reproducibility practice was designed around human-authored code, and the answers are not yet settled.