Week 5 Live Session (AI)

We gave an informal, optional introduction to using AI for R coding. We covered data privacy, the different ways to work with AI, and a live demo of having Claude write, explain and help verify code in Positron.

You'll learn:

• How to change your Claude privacy setting so your chats are not used to train models, and what 30-day retention and zero data retention agreements mean
• Why to ask AI to write code for an analysis rather than do the analysis itself, so you have a verifiable record of what was done
• How to use an AI extension in Positron to import and reshape data, make charts, and choose between manual, automatic and auto modes
• How to ask AI to add detailed comments to code and how to verify its output by running the code piece by piece and comparing it with the raw data
• Why working inside your editor with a tool that sees your R session beats copying and pasting between a chatbot and your code

Resources mentioned:

• Claude
• Claude Code
• ChatGPT
• Codex
• Posit Assistant
• Positron
• RStudio
• Privacy and AI assistants (Posit blog)
• Raycast
• Handy
• {tidyverse}
• {readxl}
• {janitor}

Key moments:

• 01:46 Participants share how they have used AI for coding
• 09:25 A participant describes using Claude for gene expression analysis
• 11:42 Ask AI to write code so results can be verified
• 14:41 Data security and the Posit privacy article
• 17:40 Turning off model training in Claude privacy settings
• 21:10 Question on enterprise accounts and organizational AI policies
• 25:10 Web app, desktop app, editor extension and terminal compared
• 30:47 Posit Assistant demo importing penguins data and making a scatter plot
• 34:33 Claude extension demo planning an Excel import
• 41:51 Dictating prompts to AI with Raycast and Handy
• 45:57 Question on how to verify AI output against raw data
• 47:41 Having AI add detailed comments to explain code
• 51:32 Question on efficiently verifying AI data cleaning, including tests