Experil: The Moral Hazard Beyond Hallucinations

Description:

We have a fundamental fallacy in AI today that we can just do better. Hallucinations are indeed improving but experil remains largely unchecked. The hazards not measured or not defined in a PERFECT model – the peril external to the model. It isn't a wrong answer, it is the unintended consequences of correct answers.

We will discuss how the inherent pace of AI is creating experil everyday. We will examine specific cases and propose new tools to identify, monitor, and mitigate these effects.