How to do a drawing on screen?

I suspect they gave it a name “AI” but it is not AI at all, but just NLP. It’s just so stupid. How could it be even called AI, when it does not have memory to remember previous interactions with humans? I mean for example, there is filemanager called fff. I got this recommendation for using fff recent days by this OpenAI. I asked it for help how to display hidden files, but it just cannot find out what is fff. It connected it somehow with Windows apps. I found help doc on github and sent it to it. It almost found it as the hotkey to be written .: the “AI” interpreted it and read it as alt+. . But reagarding writing scripts, I just hope it will be more predictible, because now it generate dozens times different result when I asked it to extract some information which it was trained on. The communication must improve and I hope, the developers could have some more official language like using some supported marks to better communicate with it using exact term like for example to structure and to describe my problem using marks similar to html or xml. And the memory must improve to memory exact conditions, arguments and rule I gave it.