Hi
Thumbs Up & Down are part of feedback loop in terms of LLM response.
====================================================================================
How does user feedback from thumbs-up or thumbs-down get used?
Thumbs-up/down feedback provides a high-level signal to help ServiceNow evaluate how generative AI models are performing across different skills and use cases. This input is used to identify general patterns or recurring issues that may inform future improvements.
However, this feedback:
Is not used for automated model retraining or fine-tuning
Does not lead to immediate, instance-specific changes
Does not guarantee resolution of any specific issue
Instead, it is reviewed by ServiceNow’s product and engineering teams to help prioritize model improvements over time—for example, identifying common points of confusion, formatting issues, or gaps in understanding that may affect multiple customers.
Feedback is only collected from customers who have opted into the Advanced AI & Data Product Terms data sharing program. When paired with usage logs and context, this feedback becomes more meaningful and helps our teams better understand and improve model behavior.
==========================================================================
How it goes in background its as below.
Have a look at This Table--> sys_gen_ai_log_metadata.
Data and Logs are too much encapsulated so not easy to decode . Have a look at table - SYS_GEN_AI_LOG_METADATA
Just tried Another Like - and 1 dislike in Chat Window and another entry appears for Both
Ane here are the entries for all the feedback.
2 like and 1 dislike
Just to Add here:-
Regards
RP
