> Since it is a big wall of text, let us ask the subject to summarize. Let's see if it did a good job.
That's the whole problem. Now you have to read the summary and the original, just to verify whether the summary was correct — especially if it's a niche topic (admittedly I'm going by the summary here).
Maybe for a certain class of low-interest work, i.e. for texts where I don't care that I would miss something interesting. For things I care about, the error rate has to come down to zero or the possibly-flawed draft is really of no use.
That's the whole problem. Now you have to read the summary and the original, just to verify whether the summary was correct — especially if it's a niche topic (admittedly I'm going by the summary here).