Exams / AIF-C01 / Transparency and explainability / Question 2
AIF-C01 practice question: Transparency and explainability (2 of 2)
Guidelines for Responsible AI A free question from our original AIF-C01 practice exam. Try it, then open the answer to see why each option is right or wrong.
A lender must give every rejected applicant the main reasons for the decision. A complex deep learning model is slightly more accurate than a simple, transparent model that shows how each input affects the result. What is the MOST appropriate choice?
Choose one answer.
- AA. Use the deep learning model, because accuracy always outweighs other factors
- BB. Use the deep learning model and tell applicants the decision was made by AI
- CC. Use a foundation model to write a plausible reason after each decision
- DD. Use the transparent model, accepting a small loss of accuracy for clear reasons
Show the answer and explanation
Answer: D
- A. Wrong: Ignoring the legal duty to explain decisions creates compliance risk, however accurate the model is.
- B. Wrong: Saying AI was used is not the same as giving the reasons for the decision.
- C. Wrong: A generated reason may not reflect how the decision was actually made, which is misleading.
- D. Right: Correct. When decisions must be explained, a transparent model that shows how inputs drive the result is often the right trade-off.
Explanation
There is often a trade-off between performance and interpretability. Transparent models, such as linear models or decision trees, show how inputs lead to outputs. Complex models can be more accurate but need post-hoc explanation tools, such as feature attributions, which only approximate their behaviour. Regulated decisions often favour the transparent option.
Study
Keep practising
Answer questions with a score at the end, or practise just this topic with the explanation after each one. No sign-up needed.