Catch the Mistake
AI makes one believable mistake. You catch it, fix it, and prove the idea.
Core trick
Prompts an LLM to inject one plausible error into a generated output, then scores whether the user spots and fixes it.
AI makes one believable mistake. You catch it, fix it, and prove the idea.
Prompts an LLM to inject one plausible error into a generated output, then scores whether the user spots and fixes it.