Exams / AIF-C01 / Responsible AI and guardrails / Question 3
AIF-C01 practice question: Responsible AI and guardrails (3 of 5)
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 company with a net zero commitment is choosing a model for a high-volume document classification task. Which TWO choices will BEST reduce the environmental impact? (Choose two.)
Choose two answers.
- AA. Choose the smallest model that meets the accuracy target
- BB. Always choose the largest model available, just in case
- CC. Use distillation to create a smaller model for the task
- DD. Retrain the model from scratch every week
- EE. Generate several answers per request and keep one
Show the answer and explanation
Answers: A, C
- A. Right: Correct. Smaller models need less compute per request, which reduces energy use.
- B. Wrong: A model larger than needed uses more compute and energy for no benefit.
- C. Right: Correct. A distilled model keeps task performance with fewer resources.
- D. Wrong: Frequent full training uses a great deal of compute and is rarely needed.
- E. Wrong: Generating extra answers multiplies compute use for each request.
Explanation
Sustainability is part of model selection. The AWS Generative AI Lens recommends smaller or task-specific models, distillation, and optimisation techniques such as quantisation and pruning to cut resource use while meeting performance goals.
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.