Supplementary file 1_Feedback-driven rule induction and retrieval-augmented bias mitigation for large language models.pdf

Most existing bias-mitigation strategies depend on fine-tuning, reinforcement learning, prompt engineering, or interventions during pre-training, requiring either parameter updates or extensive manual prompt design. These requirements limit their applicability when the underlying model is available only as a closed-source or black-box system.</p>Methods<p>This work explores whether bias-mitigation knowledge can instead be separated from the model and reused without altering its parameters.

We propose a feedback-driven external alignment memory framework for post-hoc gender bias mitigation that transforms identified bias into concise corrective rules through an iterative feedback process. These rules are stored independently of the model parameters and are retrieved during inference to influence future responses. Retrieval is performed using cosine-similarity matching between an incoming query and previously stored examples, allowing corrective knowledge to be reused while maintaining complete model independence.</p>Results<p>Evaluation on selected subsets of BBQ, BiasNLI, CoBias, CrowS-Pairs, and WinoBias shows measurable reductions in gender-related bias.