Copy link to clipboard
Copied
Hi,
ChatGPT and other LLMs are discussed everywhere. They are also used to write new code or at least improve existing code or explain code. (As I do not have that many experience with ExtendScript, I would love to get explanations for Klaus' or Rick's code ;).)
Did anyone test this already?
Can these AI systems handle ExtendScript? Or only JavaScript?
What's your experience?
Best regards, Winfried
Copy link to clipboard
Copied
I have been suprised at how "smart" these systems are. However, I have seen some "dumb" results as well. I would say to plug in some code and see what you get.
Copy link to clipboard
Copied
Does the AI even have access to a running FM, of the suitable release level, to test the code?
Does the AI have access to a data set with which to test the code?
Does the data set include limit & expected-error cases?
How was the AI trained? If the documentation for the scripting language {release level} is not on line, or is, but is incomplete or in error, expect the usual AI result of: sounds convincing & erudite, but is junk, perhaps dangerous appears-to-work, some-of-the-time, junk.
Copy link to clipboard
Copied
One a somewhat related note—in case the ChatGPT results don't pan out—did you see that Rick posted some very informative videos on automating with scripting on YouTube? (And a whole lot more.)
https://www.youtube.com/@frameexpert/videos
~Barb