This is where I collect and write things, mostly on academe and artificial intelligence. Established in 2010, by Vaishak Belle.
The qualification problem - that of specifying all the qualifiers for, say, the preconditions of actions, of which there may be very many — is not necessarily eliminated by data-driven learning. The latter only ensures everything caught in the data is captured, so eliminates the need for an expert from specifying these. However, if something unexpected were to be observed, then we’d need to address this via some expert knowledge after all.
For example, the precondition for picking up a box could be that it’s not heavy. This could be given by an expert or learnt. However, in the rare chance that we encounter a not-heavy box that is glued to the floor, then we’d need a way for the robot to understand what’s happening. Don’t matter if the axiom was learnt or declared.