Large language models (LLMs) do not have a mechanism to prevent them from giving incorrect answers. While I believe they are still useful, they will never be 100% accurate. We need to recognize that users will frequently receive wrong answers. To improve reliability, there should be a system in place that involves humans in the process of verifying the accuracy of the responses.
3 comments
[ 2.7 ms ] story [ 13.3 ms ] thread