Given my experience with 3.5, I think the model might actually help to create the code needed to turn said output into commands understood by an interface.
I am serious. I gave 3.5 some code for creating a graphic in base R and asked it to reproduce the graphic using tidyverse syntax. It didn’t create what I wanted, because the system does not understand anything, but simply predicts probable next lines of code. But I looked at the code, corrected it and got a graphic output which worked for the case in question.
I have little doubt that a task like writing code for motor control in a bloody tribble furby is within the capabilities for the model.
Now, about that…
