Identify AI concepts and capabilities

A lender uses an AI model to decide credit limits. Regulators require that when the model declines an application, the applicant can be told, in understandable terms, which factors led to that outcome. Which responsible AI principle does this requirement reflect?

• AFairness, so applicants are scored without bias against any group
• BReliability and safety, so the model keeps operating dependably
• CTransparency, so people can understand the factors behind a decisioncheck_circle Correct
• DAccountability, so a named team answers for the model's outcomes

Why A is wrong: Fairness concerns equitable treatment across groups; the requirement here is about explaining a decision, not about removing bias, so it is not the best fit.

Why B is wrong: Reliability and safety is about dependable operation; a model can be reliable yet still opaque about its reasons, so this does not meet the explainability requirement.

Why C is correct: Transparency requires that AI systems be intelligible so people can understand how outputs are produced; telling applicants which factors drove a decision is exactly this principle.

Why D is wrong: Accountability makes the organisation answerable for outcomes and is a tempting choice, but being answerable is not the same as making a decision understandable to the applicant, which is what is required.

A recruitment screening model is evaluated to confirm that it does not systematically disadvantage applicants of a particular gender or ethnic background, and that comparable candidates receive comparable outcomes. Which Microsoft responsible AI principle guides this evaluation?

• AReliability and safety
• BInclusiveness
• CFairnesscheck_circle Correct
• DTransparency

Why A is wrong: The word evaluation suggests testing that the model works, which makes reliability and safety tempting, but that principle is about consistent and safe operation, not about equitable treatment across groups.

Why B is wrong: Inclusiveness also concerns different groups of people, so it is a plausible trap, but it is about empowering and being accessible to everyone, whereas the concern here is unequal outcomes between groups.

Why C is correct: Fairness is the principle that an AI system should treat all people equitably and avoid systematically advantaging or disadvantaging particular groups.

Why D is wrong: Transparency would help explain a decision, which sounds useful in hiring, but this scenario is about the outcomes being equitable, not about the reasoning being understandable.

A team is comparing two configuration parameters on a deployed model: one bounds the length of the generated reply, while the other governs the randomness of token selection. Which statement correctly distinguishes the max tokens setting from the temperature setting?

• AMax tokens controls the randomness of the reply, while temperature limits how long the reply can be
• BBoth parameters set the reply length, differing only in the units of text they count
• CMax tokens limits how long the reply can be, while temperature controls how random token selection ischeck_circle Correct
• DBoth parameters control randomness, with max tokens applied first and temperature applied afterwards

Why A is wrong: Tempting because it names both real parameters, but it swaps their roles: length is set by max tokens and randomness by temperature.

Why B is wrong: Tempting because both are numeric settings, but only max tokens affects length, so treating temperature as a second length control is wrong.

Why C is correct: Correct: max tokens caps output length and temperature adjusts the randomness of next-token choice, so the two parameters govern different aspects of a response.

Why D is wrong: Tempting because they are often tuned together, but max tokens has nothing to do with randomness, so describing both as randomness controls is incorrect.

A retail team wants to power a semantic search feature by converting each product description into a numeric vector, so that descriptions with similar meaning sit close together and can be compared for likeness. Which model capability should they select from the Foundry catalogue for this task?

• AAn embedding model, which turns text into comparable numeric vectorscheck_circle Correct
• BA language model, which generates conversational text replies
• CAn image-generation model, which creates pictures from text prompts
• DA multimodal model, which interprets images alongside text input

Why A is correct: Correct: an embedding model maps text to dense numeric vectors whose closeness reflects semantic similarity, which is exactly what semantic search over product descriptions needs.

Why B is wrong: A language model can process the descriptions and sounds like a fit, but its capability is generating text, not producing the comparable vectors that a similarity search relies on.

Why C is wrong: Product catalogues involve imagery, which makes this tempting, but image generation produces new pictures and does nothing to represent text meaning as vectors.

Why D is wrong: A multimodal model handles mixed inputs and seems broad enough, but the task is measuring similarity between text descriptions, which calls specifically for an embedding capability rather than image interpretation.