Abstract
The abstract describes a system designed to ensure AI models make fair and ethical decisions in real-time. This system uses two separate AI "agents" working together:
- The Proposer: The first agent suggests a specific action or decision.
- The Checker: The second agent (an "agentic ethicist") reviews that suggestion to make sure it follows set rules for fairness and legality.
If the suggestion is found to be unfair or biased, a "safety layer" immediately changes it to a better, more compliant alternative before the decision is finalized. This method aims to catch and fix biased AI recommendations while they are being made, rather than trying to fix them after the fact.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Piratiyath, Heena and Kayande, Tanmay, "The Dual-Agent Architecture for Real-Time Correction of Computational Model Outputs", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10340
COinS