Marvin Minsky: Steps Toward Artificial Intelligence (1961) [pdf]

Scanning this list of topics, it seems that currently progress is being made on quite a couple of them, though maybe in a more fuzzy way than Minsky envisioned at the time of writing.

One aspect on which the progress seems to be low is planning. All the work I have seen in recent times involving planning (chess, etc) was basically using an off the shelf tree pruning planning algorithm, and the novel stuff was in using neural networks for improved cost functions. Does anyone know of explicit planning work using modern deep/recurrent neural networks?