This is where I collect and write things, mostly on academe and artificial intelligence. Established in 2010, by Vaishak Belle.
Of course, if hard-wired models are brittle because they over-rely on a human expert, data-driven models are brittle because they over-rely on the training data. We know both to be flawed, skewed /biased. We have take the assumptions underlying our models seriously before we figure out how to fix it. (And we will fix it with a hybrid of the existing two strands, presumably.)
https://blog.vaishakbelle.com/post/706599774270488576/brittleness-of-symbolic-logic