[id10t]

Depends on what you are using AI for....

I prefer to use chatgpt now for programming related searches and queries, simply because it does do a good job with that especially our old "so unsupported we can't even buy a license for it any more because IBM has no way to charge us for it" software stack (EGL/VG). At least when it gives me an incorrect answer ("how can i convert a string to a date in EGL?") I can say "no, that wasn't it" or "no, I don't have that library available" and it will correct itself. Maybe after two or three tries.

I trust it to make code for me like I would have a first semester intern do... but the AI types much faster

"Take this CSV header row and give me a java object to import it to using the OpenCSV BindByColumnName annotation, generate a DB2 create table statement, make the strings all varchar(100), add a autoincrement index field as the primary key, and generate a jpa repository to use the import object as the db entity object"

Then paste in the header row, hit go, take a gulp of coffee, and the code is done. I can do it manually with some scripting cheats I created years ago in about 45 minutes.

I always end up tweaking the field sizes for string values, fiddling with column names to match our internal standards, and depending on what number values represent I change the data type from Decimal to BigDecimal (money stuff).