Do you trust your AI? Achieving reliability on enterprise data

Reliability and trust are critical when integrating AI with enterprise data and systems. In this talk, the Hasura AI Labs team presents the AI Reliability Benchmark, a system to evaluate the reliability of a general-purpose AI assistant on enterprise data - including key insights on unreliability patterns.

The talk will cover how different AI techniques fare against the reliability benchmark and provide design patterns for building trustworthy AI systems.