SkywardAI Labs · Fairness Intelligence Labs

Fairness Intelligence Causal Approaches to Trustworthy AI

Developing causally grounded, interpretable methods for debiasing and calibrating black-box, open-weight, and agentic large language models.

Explore Research ↓
Latest 🏆 NeurIPS 2026 (CORE A*) — Why Are LLMs Confidently Wrong? Correcting Overconfident Errors via Causal Head Intervention  ·  🎉 EACL 2026 (CORE A) — Adaptive Causal Prompting for Black-Box LLM Debiasing  ·  ⭐ WWW 2026 (CORE A*) — Fairness-Aware Triggering for LLM Fairness  ·  🔬 Research in Progress: Dynamic Orchestration for Agentic AI Debiasing       Latest 🏆 NeurIPS 2026 (CORE A*) — Why Are LLMs Confidently Wrong? Correcting Overconfident Errors via Causal Head Intervention  ·  🎉 EACL 2026 (CORE A) — Adaptive Causal Prompting for Black-Box LLM Debiasing  ·  ⭐ WWW 2026 (CORE A*) — Fairness-Aware Triggering for LLM Fairness  ·  🔬 Research in Progress: Dynamic Orchestration for Agentic AI Debiasing       

Who we are

Responsible AI
Through Causal Reasoning

Fairness Intelligence is a research group at SkywardAI Labs dedicated to understanding and mitigating bias and overconfident errors in large language models using causal methods. We bridge explainable AI and causal inference to build more trustworthy, accountable AI systems.

Our research spans three tiers of model access — black-box APIs, open-weight architectures, and multi-agent agentic systems — developing unified causal principles that apply across all levels of complexity.

We publish at top-tier venues (CORE A*, A, IEEE Q1), including NeurIPS, and are committed to open, transparent research that promotes safer AI deployment in high-stakes domains such as healthcare, hiring, education, and public information systems.

8
Publications (2025–2026)
CORE A*
Top venue — NeurIPS 2026

Publications

Research Map

Four questions drive our work. Each publication links to the questions it helps answer, as a concept, a framework, or a working system.

All publications

Q1 debiasing · Q2 trust · Q3 causal mechanism · Q4 practice. Numbers are contribution weights from 1 to 3.

Team

Group Members

Four researchers, connected by the publications they have written together.