7  R and AI Resources

A curated list of resources for learning and using R, and for working with AI as part of your data-science practice. Nothing here is required reading — but each item below has earned its place. Skim it now, bookmark it, and return as you go.

7.1 Resources for learning and using R

7.1.1 IDEs for R

RStudio remains the most widely used IDE for R, and, like R, it is open-source. Its “Help” tab is a gateway to a wide range of R and RStudio resources. However, Posit (the company behind RStudio) has released Positron, a next-generation IDE built for data science with both R and Python. Positron combines the familiar feel of RStudio with the extensibility of VS Code, and includes built-in AI assistance via Posit Assistant. If you’re comfortable with RStudio, stick with it; if you’re curious about a more modern interface or plan to work across R and Python, Positron is worth exploring.

7.1.2 Learning R

The place to start is the widely used R for Data Science by Wickham and Grolemund. They will guide you through the Tidyverse. Tidy Data Tutor helps you visualize your data analysis pipelines. James Scott has a great introduction to R that maps onto BUSN 5000.

Kieran Healy’s Data Visualization book (also linked on the course schedule) is a great introduction to the wonders of ggplot, which is not R, but indispensable to analysis done in R.

Learning R Markdown or Quarto should go hand-in-hand with learning R. For R Markdown, get started here and keep a cheatsheet handy. You can get started with Quarto here.

7.1.3 Workflow

At least as important as learning R is understanding basic workflow principles. R-bloggers has a beginner’s guide using RStudio Projects. You can find some of the same principles described in chapter 2 of Healy’s book. Matt Gentzkow and Jesse Shapiro provide a fuller take on workflow principles from the perspective of academic social scientists (also posted on eLC). They even provide a manual for doing as they do.

For a perspective specifically on coding practices in data science education, Pruim, Gîrjău, and Horton (2023) make the case in “Fostering Better Coding Practices for Data Scientists” (Harvard Data Science Review) that good coding habits matter even for beginners. Their argument: it’s far easier to learn good practices alongside R than to unlearn bad habits later. They organize their guidance into a practical top-10 list. Worth reading early in your data-science career.

7.2 AI in August 2026

7.2.1 The main AI systems and keeping up with them

In August 2026, widely used general-purpose AI products include ChatGPT from OpenAI, Claude from Anthropic, and Gemini from Google. These products are powered by underlying models, and both the products and models change frequently.

If your experience has been exclusively through a chat interface, you should know that AI tools can now carry out longer, multi-step tasks. This shift is reflected in the rapid growth of coding agents such as Codex and Claude Code. Availability and usage limits vary by product, surface, and plan; Codex currently has an entry point on the ChatGPT Free plan as well as higher-usage paid plans.

The AI landscape is changing fast. To keep up, I recommend following Ethan Mollick’s Substack. If you want to swim in the deeper end of the pool, check out Zvi Moshowitz’s Substack, Don’t Worry About the Vase.

To learn more about AI and how to use it, check out the courses offered by OpenAI Academy and Anthropic Academy.

7.2.2 Coding with AI

One place AI really shines is coding. Chat assistants can answer a question or explain an error. An agentic coding tool can work across an entire project: it reads the relevant files, edits and runs code, checks the result, and reports what it changed. You describe the goal in plain English and remain responsible for reviewing the work. Whether the tool asks before each edit or proceeds automatically depends on the permission mode you select.

The landscape moves fast; as of mid-2026, the main options fall into two groups.

Built into your IDE. The assistant works right where you write your code:

  • Posit Assistant: Built into Positron as its default AI experience. It can use context from your live data-science session, including loaded data, plots, and console history. You configure one of its supported model providers.
  • GitHub Copilot: The most widely available assistant. It works in VS Code and Positron and, as an opt-in, in RStudio Desktop. It suggests code as you type and can also work as an agent. Verified students can often use it free through GitHub Education.

Standalone coding experiences. These work with a whole project rather than living inside the editor where you type R code:

  • Claude Code: Anthropic’s agentic coding tool, available in desktop and web experiences. Open your project and describe the task. If you want to review a proposed approach before any edits, select Plan mode.
  • OpenAI Codex: OpenAI’s agentic coding experience in the ChatGPT desktop app. It can work directly with a local project, and its permission settings determine what it may do without stopping for approval.

Finally, Cursor is an entire code editor rebuilt around AI. Software developers have embraced it, but for R work you would give up the data-science conveniences of RStudio and Positron.