If you aren't able to reverse engineer quillbot technology, it's low scale.
There is no way to use it with API. Unless you wish to buy 10 licenses to produce 10 arguably detectable articles per minute.
Do I always shit on quillbot? I think I said this is a good tool on a number of occasions.
Anyways, it's quite limited for the purpose.
As a tidbit, I ran a lot of quillbot content through duplicate checkers using w-shingling and jaccard indexes and the results were anywhere between 5 - 20% duplicate shingles comparing to the original piece.
The setting was creative / expand + 3rd level of synonyms.
There were shingles where in a 1000 word long article there were 2 or 3 identical sentences in a single shingle.
It cannot change the sentence structure everywhere, leaving some evident footprint.
This is too much to pass
the most basic content plagiarism checks.
I won't even mention technology at the disposal of Google.
That's irrelevant and my bet is you would probably say it's the otherwise despite the evidence in patents.
Keep in mind a lot of big tech companies started to heavily use machine learning in the past decade.
If you forget it, you're pretty much living under a rock.
Modern SEO is about big data and machine learning. Anything else is s*it you might use to become a virtual assistant or manual SEO worker in corporate world.
