A web toy scores how predictable your writing is to AI
Users write 8 short text . The service checks how easy each word would be for AI models to predict. More predictable words mean the writing looks more like text.
The score uses logprobs for each word. It currently uses including Llama 3.1, , and Qwen3 to keep costs low. The scoring is tuned so ChatGPT and Claude answers land near 100%, while more interesting human answers land around 10% to 30%.
It is free, needs no signup, and lets users share results with friends.
Key points
- Users write 8 short and get a predictability score.
- More predictable wording is treated as more AI-like.
- The score is based on logprobs for each word.
- The site uses Llama 3.1, , and Qwen3 to control costs.
- It is free, requires no signup, and supports sharing results.