A lot of people seem to be under the impression that combining GOFAI and contemporary AI will make models more general. I'm particularly interested in reasoning through analogy or case-based reasoning.
Combining GOFAI with New AI is usually described as a grounding problem. That means, to verify if the AI is really intelligent. In the literature this concept is called “Total turing test”, which means that a score is calculated which gives the similarity to a human. IMHO it is a bit difficult to identify last papers about the subject, because reaching a high similarity is equal that somebody has build a human-level-ai robot, for example the android Data of Startrek would fulfil the requirements. But also in real life applications some challenges are done, for example the Jeopardy challenge, or a robot who must dance like a human actor.
Google Scholar finds 3 paper from the last year which are going into the right direction.