I found this paper on endowing AI with common-sense knowledge readable, novel and persuasive. https://www.sciencedirect.com/science/article/pii/S1364661320302163

There's been a quite of bit of good work starting from simpler models. Rod Brooks and Pattie Maes made a series of robots in the late 80s and early 90s that were much closer to "intelligent" than most of what's been done in recent years. http://www.ai.mit.edu/projects/hannibal/hannibal.html

@dnmfarrell nice writeup. If you liked that I think you’ll like Josh Tenenbaum’s work: .