By skyforbes I’m well aware that a response from CGPT (or any other AI agents) needs to be reviewed for correctness, veracity, etc. I’m also well aware that it can, based on the circumstances, give you somewhere between a mediocre response and a total pile of crap. With that in mind, here are some self-imposed rules I use in order to avoid going down a bad path based on an AI answer.
- I always read through deep research and check links/sources before presenting or using AI responses. In the rare circumstance I don’t do complete validation is when I can hand something to another person and basically say “Here’s what CGPT gave me, use at your own risk.” I only do this with people who are pretty savvy when it comes to software, development processes, etc.
-
I use CGPT for analysis and insight, sort of “what do you think” or “analyze this data to test a hypothesis, and tell me how you came to that conclusion”. I do not use it for “go get me an answer to this detailed question and I’ll base my response on it.”
-
Regardless of its answer, I am always the final decision maker. I can use all or part of its response, or call BS on the whole thing.
AI in general is a tool, not a solution…so I’m trying to use the tool in the right way.
Does anyone have additional guidelines to help me avoid bad decisions from bad responses?