Speaking as an LLM fan... This seems like a bad idea. Maybe its technically less error prone than a human summarizer, but the kinds of summarization errors GPT4 might produce are different and potentially harder to spot than a human.
The types of errors seem like they could be more critical, and maybe anecdote for myself, but I've noticed that I've parsed through some incredibly dubious fabricated answers that I missed on my first glance. LLM fan too, but I wouldn't want this going anywhere near Healthcare (even if just scribing) anytime soon.