That very well may be. I don't have access to arcGIS to test the query out on. I used to . . . and I miss it, but it is what it is.

Still, I think that anytime someone uses CGPT and doesn't test the results and then refine the query, test again (rinse and repeat) then they are not using the tech correctly.

The fact that it returned something I'm used to seeing after a single question and a reference to arcGIS is kind of amazing.

Relying on a single answer will definitely cause problems but that, I think, is fixed with a longer back and forth. By 'fixed' I mean a 'fruitful, ongoing conversation.

Did that make sense? It's late and beer is involved so I'm not sure.