You may see an “AI-resistant” tag on certain questions. This means that the question was designed to be resistant to AI language models, such as ChatGPT, and tested to ensure that they are unable to answer the question correctly.

As AI tools get updated over time, each question with an “AI-resistant” tag is periodically re-tested to ensure that it remains resistant and unlikely to be solved by AI.

As language models can vary their responses, each AI-resistant question is tested multiple times during a testing period, and is only considered AI-resistant if the AI is unable to score higher than 42% on the question, on average.

There is still a small chance that an AI will be able to solve a question, even if it is AI-resistant. This is why it’s best to have multiple AI-resistant questions in a test, as then the chance of AI passing the test becomes extremely low.

To ensure the best possible protection, we recommend combining AI-resistant questions with other forms of cheating protection, such as webcam and screen proctoring, or our live interview environment.