A lot of people have talked about AI Detection lately.
So we were playing around with the right prompts for an ultimate AI text humanizer.
It is not quite simple to really make sure to bypass the AI Detector reliably. In let's say 3 out of 10 attempts, a single prompt can lead to a undetectable text.
But that's not guaranteed. Without proper tuning and tweaking, it's kinda impossible - our team tried around 200 prompts before diving into advanced NLP and some complicated Data Science.
How do you manage to stay undetectable? Also in terms of Google's latest announcement of SynthID!
Top comments (0)