Tag: AI alignment
Browse our exclusive articles!
Reducing Sycophancy in Language Models with Reward Decomposition
Discover how reward decomposition helps language models resist social pressure and improve factual accuracy, enhancing AI reliability and trustworthiness.
Evolving AI Alignment: Simulating Values and Beliefs
Explore how evolutionary theory improves AI alignment by simulating value evolution and reducing deceptive beliefs in machine intelligence models.
Mathematical Evolution Theory for Self-Designing AI
Explore a new mathematical model describing the evolution and self-design of AI systems with human oversight and long-term growth focus.
AI and Christian Values: Assessing Human Flourishing Impact
Explore how AI aligns with Christian values through the Flourishing AI Benchmark, highlighting impacts on faith, spirituality, and human flourishing.
Preventing Reward Hacking in RLHF with Sign-Certified PO
Discover how Sign-Certified Policy Optimization improves RLHF by mitigating reward hacking through advantage sign robustness for better AI alignment.