Author

Tejaswi V. Panchagnula

h-index 0 View on Semantic Scholar

We have 1 of 1 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Preprint Jul 2026

Inference-Time Mitigation of Adversarial Political Bias in Large Language Models

The proposed Recursive Self-Correction approach raises model performance from a Political Neutrality Likert scale baseline of 2.14 to 4.56, averaged across all models, demonstrating effective inference-time mitigation of political bias in LLM-generated summaries.

Tejaswi V. Panchagnula, Bruce Coburn, Bryce J. Dietrich et al. · 0 citations