The difference between quality in GPT-3.5 to GPT-4 is game changing. It's hard to explain without you trying it for yourself.
If it's ChatGPT I'll offer to run the prompt through GPT-4 for them. It's better 100% of the time, and it negates their complaint probably 90% of the time.
I know it's not perfect, but GPT-4 is categorically different from ChatGPT.
But 4 did, and could describe the scene. And it even appreciated its relation to our discussion on epistemology when I said that I sometimes "hear" its voice when reading its replies!