I have been experimenting extensively with ChatGPT. To get it to be better at producing quality Xojo code requires a pre-prompt. I’ve been slowing building one with the hope that eventually we make build into the iDE some kind of access to a LLM that is tuned for helping users learn Xojo and write code. More experimentation will be required before I can make a determination as to whether or not it can be sufficiently tune to produce the level of quality we would require.
ChatGPT and others periodically retrain their systems but that’s very expensive so they don’t do it often.