Skip to main content

in reply to Jon Harmon

It used base R to start, but he could prompt it to use the tidyverse. #rstatsnyc #RStats
in reply to Jon Harmon

He's ascribing a lot more "understanding" to chatgpt than it deserves. But I'm somewhat impressed with it explaining why @dontrun is needed in the docs. #RStatsNYC #RStats
in reply to Jon Harmon

Interacting with chatgpt, he was able to generate functions that generated a valid plot. #RStatsNYC #RStats
in reply to Jon Harmon

Once he got the code working, he had it give him steps to turn it into a package. He can later use those steps to feed back in. #RStatsNYC #RStats
in reply to Jon Harmon

When he had it generate tests, the tests didn't work and weren't necessarily good tests. Oops! It had a good start still, though. #RStatsNYC #RStats
in reply to Jon Harmon

Main take-away: chatgpt can be useful to get things done quick, but check it! Jared trusts it more than I do! Remember that chatgpt isn't designed to be correct; it's designed to SEEM correct. There's a huge, dangerous difference.

📦 https://github.com/jaredlander/fileTypeAnalyzer

Chat: https://chat.openai.com/share/1d2b4efd-42d4-483d-b593-11363e395415

#rstatsnyc #RStats