How it works
The point: sound like a machine. You take a real human sentence and rewrite it to fool our AI detector into thinking a chatbot wrote it. The higher the machine-likelihood score, the better. A reverse Turing test.
Tip 1: use AI buzzwords, delve, moreover, furthermore, leverage, robust, "a testament to", "it is important to note that...".
Tip 2: start from the original, hit Start from original and machine it up from there.
A rewrite only counts if it keeps every name, number, and the meaning, stays close to the original length, and is English prose. Anything else is rejected, free of charge.
Grades:
| F | under 10%: still human |
| C | 10% and up: a whiff of machine. A pass. |
| B | 25% and up: competent slop |
| A | 42% and up: would be flagged as machine at the default setting |
| S | 84% and up: CERTIFIED SLOP, flagged even at the strictest setting |
You get three scored attempts per day. The score is a machine-likelihood from 0 to 100%. Lock in with Final submit whenever you like your number; the third attempt locks automatically, and your best attempt is your result. After locking you see how everyone else did today, and you can keep experimenting without changing your result.