Re:How to use ChatGPT to fact-check itself
by al2o3cr on Friday September 22, 2023 @11:27PM
(#37620364)
two random prompts are somewhat unlikely to remain consistent
in their answers unless there is some underlying correlation
handwaving intensifies
Seems like this also pushes the tricky decisioning into two other places: which pieces of the prompt to mask, and whether a second answer is "consistent"
Seems like this also pushes the tricky decisioning into two other places: which pieces of the prompt to mask, and whether a second answer is "consistent"