Responsible AI practicesFull notesSummaryRingkasanStoriesPracticeResponsible Considerations to Select a Model10 exam-style questions on this lesson.Question 1 of 10Why does the course say "face recognition is a technology, not a use case"?AFace recognition is illegal in every countryBFace recognition accuracy can't be evaluatedCDifferent face recognition uses need different tuningDFace recognition only works on live videoCheck answerQuestion 2 of 10A police department uses face matching to search for a missing child in a gallery of photos. Missing the child is far worse than reviewing extra false matches. Which metric should the system favor?APrecisionBLatencyCPerplexityDRecallCheck answerQuestion 3 of 10An online exam platform uses face recognition for virtual proctoring. Flagging innocent students is disruptive and not useful. Which metric should it favor?APrecisionBContext windowCModel sizeDRecallCheck answerQuestion 4 of 10An online store uses generative AI to persuade a specific customer group to buy. Compared to writing neutral catalog descriptions, which risks increase?AOnly veracityBVeracity, unwanted bias, and toxicityCNone, persuasion carries no extra riskDOnly latencyCheck answerQuestion 5 of 10A team compares two models on a benchmark and picks the winner, then deploys it on very different customer data. What principle did they overlook?ABigger models always perform better on any dataBLicensing never restricts how a model is usedCPerformance depends on both the model and the test datasetDLatency doesn't matter when choosing a modelCheck answerQuestion 6 of 10A startup wants to sell a product built on an open model. Which performance factor must it check before launch?AContext window, because it limits prompt lengthBNumber of hidden layers, because it sets accuracyCTemperature, because it controls commercial outputDLicensing, because some licenses restrict commercial useCheck answerQuestion 7 of 10A team needs to pass entire 200-page contracts to the model in one prompt. Which factor matters most?AContext windowBInference optionsCSustainabilityDLicensingCheck answerQuestion 8 of 10. Select two.Which TWO are part of responsible agency when selecting a model? (Select TWO.)AUnlimited autonomy without oversightBValue alignmentCHiding the model's reasoningDIgnoring accountabilityEAn appropriate level of autonomy with human oversightCheck answer0 of 2 selectedQuestion 9 of 10A company evaluates the energy use and GPU demands of different models before choosing one. Which consideration is this?ALatencyBEnvironmentalCEconomicDLicensingCheck answerQuestion 10 of 10A board weighs the efficiency gains of automating a department against job displacement and concentration of power in a few companies. Which consideration is this?AContext windowBEnvironmentalCEconomicDModel sizeCheck answerAmazon Services and Tools for Responsible AI12 exam-style questions on this lesson.Responsible Preparation for Datasets7 exam-style questions on this lesson.