1AI Tools for Scientists¶
1.1Who this is for¶
Researchers who can already program a little and who want to use these tools for real work: literature triage, data extraction from documents, code they will actually ship, drafts they will actually submit. We assume no machine-learning background. Every chapter is a runnable notebook — the outputs you see were produced by the code above them.
1.2How to read it¶
Chapters are short and independent enough to read out of order, but the four parts are ordered deliberately:
Foundations — what a model is doing when it answers, why it is fluent and confidently wrong at the same time, and when a local model is the right call.
Retrieval and Context — embeddings, semantic search, and retrieval-augmented generation, which is how you get a model to answer about your corpus.
Doing Work With Models — agents, code, analysis, writing, and structured extraction, each with the failure modes that matter for research use.
Trusting the Output — reliability, evaluation, and limitations. This is the part a statistician will find most familiar and most neglected in practice.
1.3Companion textbook¶
This site expands the AI-tools material from An Introduction to AI for Statisticians
in Python (Reid & Caffo), whose LaTeX source lives in the book/ directory of the
same repository. Where the textbook is compressed, this site is discursive and
executable.