Shiny things 1/?
post openai used their new algo to explain the behaviour of neurons in their older algo.
Explainability is hard; I think its more likely that credible-sounding explanations for AI decisionmaking will be a big business in the short term. It will be harder for regulators & the public (or anyone on the losing end of a decision) to challenge these explanations. how will we define "sufficient explanations?" via ben evans.
We need a standard measure for risk: the "micromort" via jan chipchase
new yorker goes thru the wirecard saga, which I first heard on invisibilia podcast. (RIP Invisibilia). Truly crazy.
tokyo joe via twitter - this is incredible stuff.
Don't miss what's next. Subscribe to Magpie Chronicles: