Traceability in GPT-3.5 Generated Response

Hey everyone, Anyone know how one can implement traceability in GPT-3.5 solutions where trace the generated output is aligning with the user input or not. Instead of human feedback, is there any technique or mechanism for traceability of generated results with user input?