Video lectures

Here are three interesting video lectures, I hoped to post more but I haven’t found any more recently.

Eliezer Yudkowsky: The Challenge of Friendly AI—This talk introduces to the concept of Friendly AI by starting with some specific AI (chess player), going meta enough times that a Reflective AI becomes a possibility, then Eliezer discusses the problem of showing that a self modifying agents wont lose its morals.

Judea Pearl Tribute Symposium: CausalityA mathematical model of causality is sketched then its connections with probability and counterfactuals and belief propagation are mentioned and explained by example.

Shane Legg: Universal measures of intelligence—Discussion about the research and experiments on a particular computable formula Algorithmic Intelligence Quota that aims to be a step towards making the notion of intelligence scientific.