Bharathi Srinivasan

AI agents are increasingly making real-world decisions — approving loans, issuing refunds, offering discounts — yet most fairness testing still focuses on whether chatbots use offensive language. This gap creates real business and regulatory risk.

Discover how Automated Reasoning checks in Amazon Bedrock Guardrails enhance the trustworthiness of AI applications. In this workshop, we will showcase the synergy between automated reasoning and generative AI to create highly reliable systems, delivering up to 99% verification accuracy.

Implementing guardrails for large language models (LLMs) involves balancing performance with responsible AI practices. This interactive chalk talk explores approaches for designing safety filters and use case-specific controls while optimizing for accuracy, latency, and cost.

Large language models (LLMs) are improving rapidly at generating correct answers, but they are still susceptible to hallucinations. Hallucinations are responses that are factually incorrect, have unreliable information, or are inconsistent with the training data.