Is there any work being done on summarization based on loading content in and doing some sort of manipulation on the internal representation? It seems to me tha

Is there any work being done on summarization based on loading content in and doing some sort of manipulation on the internal representation? It seems to me that this ought to be a better solution for summarization than simply asking the model to summarize and hoping its training set taught it how to do what you want. Seems like it would generally fix the potential confabulation issues as well w.r.t. summarization at least.