Abstract
Proposed herein is a system that chooses how much testing is needed for an improvement to an artificial intelligence (AI) root cause analysis (RCA) agent before the system trusts the improvement. Small, local changes can receive quick replay first, while broad or risky changes require a full rerun or multi-case batch testing. One result of the proposed approach is a safer and less costly self-improvement loop through which a system can spend additional testing effort when the change warrants such testing.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Cehreli, Necati, "TOKEN-EFFICIENT VERIFICATION FOR SELF-IMPROVING ROOT-CAUSE-ANALYSIS AGENTS", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/11358