AI How to Part I: Using GenAI Tools in Your Browser

An easy, low-effort way to get a feel for GenAI tools is through browser-based interfaces. They do not require any installation or setup and do not have a steep learning curve to get started. Below is an example of using Claude to modify an R script to change figure aesthetics. Tweaking figures is one of my favorite ways to use GenAI tools. Results are easy to verify and I often find it’s much faster than looking up documentation or searching through Stack Overflow.

Browser based

Note the original code is from the R graph gallery

“You are an expert R programmer. Modify the following R code to use red colors in p, blues in p1, greens in p2 and purples in p3. Use cowplot to make a multipanel figure with the title ‘AI tools are pretty cool’”.

# library

library(ggplot2)
library(ggExtra)

# The mtcars dataset is proposed in R

head(mtcars)

# classic plot :

p <- ggplot(mtcars, aes(x=wt, y=mpg, color=cyl, size=cyl)) + geom_point() + theme(legend.position="none")

# with marginal histogram

p1 <- ggMarginal(p, type="histogram")

# marginal density

p2 <- ggMarginal(p, type="density")

# marginal boxplot

p3 <- ggMarginal(p, type="boxplot")

Activity: Try it out!

Select an AI interface to further modify your plot. You can use any of the following: ChatGPT, Claude, Gemini, or another of your choosing. You will probably need to sign up for an account if you haven’t already.
Use a prompt such as: “You are an expert R programmer. Modify the following R code to {use your imagination here}. Copy the R code from the”Prompt” window above, and paste it into the chat window. Or use your own code for an existing figure and experiment with a prompt you think would improve the figure aesthetics.
Inspect the results and if they seem reasonable, copy and paste the generated code back into your R session and run it.

How did it work?

Congratulations, you are have now dipped your toes into the world of vibe coding!