Open Assessment · RAIF
Responsible AI Foundation practice test
20 free questions across the six RAIF domains, with the correct answer and the reasoning explained. Open all year, as many times as you like. It grants no certificate or badge: it is preparation for the official RAIF exam.
Start the assessmentFree
No account, no card, no limit on attempts.
Not the certification exam
Separate questions, never reused in the exam.
Nothing on your record
Your answers are not kept. Only the result by domain is saved, linked to you only if you are signed in, and it produces no credential.
Sample questions
Two questions from this practice test, in the same style as the certification exam. Answer before opening the explanation.
Question 1
A system reads each incoming email and labels it as "invoice", "complaint" or "other". What kind of AI task is this?
- Content generation, because the system writes a new label for every email
- Classification, because the system assigns each input to one of a set of categories
- Robotics, because the system acts automatically without a person
- Translation, because the system converts emails into a different format
Show answer and explanation
Correct answer: B. Classification, because the system assigns each input to one of a set of categories
Sorting inputs into predefined categories is classification, one of the most common uses of AI at work. Choosing a label from a fixed list is not generating new content. Robotics involves physical machines acting in the world. Translation converts content from one language or form to another, which is not what happens here.
Question 2
In the context of generative AI tools, what is a "prompt"?
- A warning the tool shows when it is not sure about an answer
- The data set the tool was originally trained on
- The licence that allows a company to use the tool
- The instruction or input a user gives the tool to get a response
Show answer and explanation
Correct answer: D. The instruction or input a user gives the tool to get a response
A prompt is what the user types or provides, such as a question, an instruction or a document, to get the tool to respond. Uncertainty warnings, when they exist, are part of the output, not the prompt. Training data is what the model learned from before release. A licence is a contractual matter and has nothing to do with prompts.