Agree with the other commenter the new approaches are mostly neurosymbolic and based on DNNs vs the approaches given here. FWIW I think this is one of the critical pieces to AGI and I know many of the large AI labs are working on integrating these approaches in an agentic way with the LLMs they’ve trained. I expect this will dominate at least the next 12-24 months of research. There are also a few AI startups out there also working on prototyping these ideas.