Later, I had a Java programming problem to solve with Google protobufs. My co-workers weren't coughing up any answers. I handed it off to ChatGPT4. The answer was reasonable, but not 100%. I resolved it on my own later this morning.
The insight was interesting, but GPT4 didn't bowl me over with a great solution for either case. I have a challenge to be the integrator of a large sensor network, and AI/ML evaluation platform and a hazard prediction platform. Everyone has their own way to do things, so I'm obliged to accommodate each one and act to translate between platforms to get the desired result. The key target is SPEED. The intent is to demonstrate an evaluation of data using AI/ML faster than the expert humans and deliver it to the troops in the field in near real time. Demo in 3 weeks. Lots to do.
GPT5 will be much smarter, then GPT6, then GPT 7, then …
It appears that a Great Solution MUST be out there in the cloud for AI to find to give to you.
(or a bunch of WRONG answers to test)