Unit 4: Practical Prompt Design with Vertex AI - Practice Quiz

CSG202 — Generative Ai Fundamentals 60 Questions
0 Correct 0 Wrong 60 Left
0/60

1 What is a key business reason for adopting responsible AI practices?

Explain the business case for responsible AI Easy
A. Removing all human oversight from decisions
B. Eliminating the need for data entirely
C. Guaranteeing 100% model accuracy
D. Building customer trust and brand reputation

2 Which of the following is a potential risk of NOT practicing responsible AI?

Explain the business case for responsible AI Easy
A. Reputational and legal harm
B. Faster model training
C. Lower cloud storage costs
D. Improved data compression

3 Responsible AI is best described as an approach that considers what throughout the AI lifecycle?

Explain the business case for responsible AI Easy
A. Only the speed of computation
B. Ethical, social, and legal implications
C. Only marketing outcomes
D. Only hardware requirements

4 How can responsible AI provide a competitive advantage for a company?

Explain the business case for responsible AI Easy
A. By reducing product features
B. By avoiding all use of AI
C. By ignoring customer feedback
D. By differentiating the brand as trustworthy

5 Which stakeholder group benefits from an organization's responsible AI efforts?

Explain the business case for responsible AI Easy
A. Only competitors
B. Only software vendors
C. Customers, employees, and society
D. Only the finance department

6 What does "issue spotting" in AI ethics primarily involve?

Identify ethical considerations with AI using issue spotting best practices Easy
A. Choosing the cheapest hardware
B. Increasing model training speed
C. Identifying potential ethical problems early
D. Reducing the number of data features

7 Which of the following is a common ethical concern in AI systems?

Identify ethical considerations with AI using issue spotting best practices Easy
A. Bias in model outputs
B. Colorful user interfaces
C. Faster GPU performance
D. Smaller file sizes

8 Why is fairness an important ethical consideration in AI?

Identify ethical considerations with AI using issue spotting best practices Easy
A. It helps avoid discriminatory outcomes
B. It increases file storage
C. It reduces electricity usage
D. It speeds up code compilation

9 Which practice supports privacy as an ethical consideration in AI?

Identify ethical considerations with AI using issue spotting best practices Easy
A. Protecting personal data from misuse
B. Storing passwords in plain text
C. Ignoring user consent
D. Sharing all data publicly

10 Transparency in AI refers to what?

Identify ethical considerations with AI using issue spotting best practices Easy
A. Using only closed-source tools
B. Making outputs random
C. Hiding model details from users
D. Being clear about how AI systems work

11 In what year did Google first publish its AI Principles?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Easy
A. 2018
B. 2010
C. 2005
D. 2022

12 Which of the following is one of Google's stated AI Principles?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Easy
A. Be socially beneficial
B. Maximize profit above all
C. Prioritize speed over safety
D. Avoid documentation

13 Google's AI Principles also include a list of application areas it will NOT pursue. What is one such area?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Easy
A. Language translation tools
B. Photo organization apps
C. Cloud storage services
D. Technologies that cause overall harm

14 According to Google, AI should avoid creating or reinforcing what?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Easy
A. Data backups
B. Unfair bias
C. Software updates
D. New user accounts

15 A key lesson Google shares from operationalizing AI Principles is that responsible AI requires what?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Easy
A. No human involvement
B. Ongoing governance and review processes
C. A single one-time checklist
D. Only external audits

16 What does it mean to "operationalize" responsible AI?

Adopt a framework for how to operationalize responsible AI in your organization Easy
A. Turn principles into concrete practices and processes
B. Delete all AI models
C. Stop using AI completely
D. Publish principles without acting on them

17 Which element is important when building a responsible AI framework in an organization?

Adopt a framework for how to operationalize responsible AI in your organization Easy
A. Clear roles and accountability
B. Avoiding stakeholder input
C. Removing all documentation
D. Eliminating review steps

18 Why is executive leadership support important for responsible AI?

Adopt a framework for how to operationalize responsible AI in your organization Easy
A. It slows all AI projects down
B. It replaces the need for engineers
C. It removes the need for ethics
D. It drives organization-wide commitment and resources

19 Which activity helps operationalize responsible AI in practice?

Adopt a framework for how to operationalize responsible AI in your organization Easy
A. Avoiding documentation
B. Conducting regular risk assessments
C. Skipping employee training
D. Ignoring model performance

20 Employee training on responsible AI primarily helps to achieve what?

Adopt a framework for how to operationalize responsible AI in your organization Easy
A. Faster internet connections
B. Lower software licensing costs
C. A shared understanding of ethical practices
D. Reduced office space needs

21 A retail company is deciding whether to invest in responsible AI practices before launching a customer-facing recommendation engine. Which argument best represents the business case for doing so?

Explain the business case for responsible AI Medium
A. Responsible AI eliminates the need for any human oversight of the system
B. Responsible AI removes the requirement to collect training data
C. Responsible AI guarantees the model will always achieve higher accuracy than competitors
D. Responsible AI builds customer trust and reduces reputational and regulatory risk, protecting long-term revenue

22 An executive claims that responsible AI is "just a cost with no return." Which counterpoint most directly addresses the financial value of responsible AI?

Explain the business case for responsible AI Medium
A. Avoiding biased or harmful outcomes prevents costly recalls, lawsuits, and lost customers
B. It transfers all liability to the cloud provider
C. It slows product development, which reduces spending
D. It ensures the company never has to update its models again

23 A startup wants to attract enterprise clients in regulated industries. How does a strong responsible AI posture support this business goal?

Explain the business case for responsible AI Medium
A. It signals accountability and compliance readiness, easing procurement and due diligence
B. It guarantees government contracts automatically
C. It replaces the need for security certifications
D. It allows the startup to skip data privacy laws

24 Which scenario best illustrates responsible AI creating competitive advantage rather than merely avoiding harm?

Explain the business case for responsible AI Medium
A. A firm delays every product to avoid any decisions
B. A firm hides its model behavior from all users to reduce complaints
C. A firm markets transparent, fair lending decisions and wins customers who distrust competitors
D. A firm stops measuring model performance to save money

25 A hiring model is trained on 10 years of past hiring decisions at a company where most engineers hired were men. What ethical issue should you spot first?

Identify ethical considerations with AI using issue spotting best practices Medium
A. Excessive inference latency
B. Historical bias in training data that can perpetuate discrimination
C. Overfitting to the validation set
D. Insufficient GPU memory during training

26 Using issue-spotting best practices, which question is most useful when reviewing a new AI feature for ethical risk?

Identify ethical considerations with AI using issue spotting best practices Medium
A. What font should the UI use for results?
B. Who could be harmed by errors, and are impacts distributed unfairly across groups?
C. Which cloud region is cheapest for hosting?
D. How many API calls per second can the system handle?

27 A medical chatbot confidently gives incorrect dosage advice. Which pair of ethical considerations is most directly implicated?

Identify ethical considerations with AI using issue spotting best practices Medium
A. Latency and throughput
B. Branding and marketing
C. Storage cost and scaling
D. Safety and reliability

28 During a design review, a team notices their facial analysis model performs far worse for certain skin tones. This is an example of spotting which issue?

Identify ethical considerations with AI using issue spotting best practices Medium
A. Prompt injection vulnerability
B. Fairness disparity across demographic groups
C. Data drift over time
D. Vendor lock-in risk

29 Which practice best supports proactive issue spotting rather than reacting after deployment?

Identify ethical considerations with AI using issue spotting best practices Medium
A. Waiting for user complaints to reveal problems
B. Conducting structured harms and fairness assessments during design
C. Disabling logging to protect performance
D. Only reviewing model accuracy at launch

30 Google's AI Principles include both things AI applications should do and areas Google will not pursue. Which of the following is an example of the latter?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Medium
A. Incorporate privacy design principles
B. Be socially beneficial
C. Be built and tested for safety
D. Technologies that cause or are likely to cause overall harm

31 A key lesson from how Google operationalized its AI Principles is that principles alone are insufficient. What did Google add to make them actionable?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Medium
A. Review processes, governance structures, and tooling to apply principles to real projects
B. A ban on all new AI research
C. A one-time employee memo
D. Outsourcing all ethics decisions to customers

32 Which statement best reflects a lesson Google learned about applying AI Principles across many products?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Medium
A. Fairness only matters for external products
B. Ethical judgment can be fully automated with a checklist
C. Principles should be kept secret from developers
D. Real cases are nuanced and often require case-by-case review and escalation

33 Google's AI Principle to "avoid creating or reinforcing unfair bias" is put into practice primarily through which activity?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Medium
A. Removing all documentation to simplify code
B. Increasing model size until accuracy peaks
C. Testing datasets and models for representational and fairness issues
D. Marketing the product as unbiased

34 Why did Google emphasize that its AI Principles are meant to "evolve over time"?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Medium
A. Because competitors publish new principles each year
B. Because they were written incorrectly at first
C. Because AI technology, use cases, and societal expectations keep changing
D. Because principles are legally required to change annually

35 Your organization wants to move from responsible AI values to daily practice. Which first step best operationalizes those values?

Adopt a framework for how to operationalize responsible AI in your organization Medium
A. Define clear roles, review processes, and accountability for AI decisions
B. Announce the values publicly and consider the work complete
C. Buy the largest available model
D. Restrict responsible AI to the legal team only

36 A framework for operationalizing responsible AI recommends embedding checkpoints throughout the ML lifecycle. Where should the first fairness and harms review ideally occur?

Adopt a framework for how to operationalize responsible AI in your organization Medium
A. During the final marketing review
B. Only when a complaint is filed
C. During problem definition and data planning, before building
D. Only after the model is deployed to users

37 Which combination best represents the organizational pillars needed to sustain responsible AI over time?

Adopt a framework for how to operationalize responsible AI in your organization Medium
A. Only a published policy document
B. Only external auditors
C. Only advanced hardware
D. People, processes, and technology working together

38 A team builds a strong responsible AI policy but ignores culture and incentives. What is the most likely operational outcome?

Adopt a framework for how to operationalize responsible AI in your organization Medium
A. Models will automatically become fairer
B. Governance reviews become unnecessary
C. Employees may bypass the policy because behavior is not reinforced
D. The policy will be followed perfectly regardless of incentives

39 To operationalize responsible AI, an organization creates an escalation path for ambiguous ethical cases. What is the main purpose of this path?

Adopt a framework for how to operationalize responsible AI in your organization Medium
A. To eliminate the need for developer judgment
B. To route hard, high-risk decisions to appropriate experts and leadership for review
C. To slow all projects equally
D. To hide difficult cases from leadership

40 Which metric approach best supports ongoing accountability once a responsible AI framework is in place?

Adopt a framework for how to operationalize responsible AI in your organization Medium
A. Rely solely on user complaints as the metric
B. Track only revenue impact
C. Continuously monitor performance and fairness metrics and report them to stakeholders
D. Measure fairness and safety only once at launch

41 A retail company is deciding whether to invest in responsible AI practices before launching a recommendation engine. Which argument best captures the strategic business case, rather than a purely ethical or compliance-driven one?

Explain the business case for responsible AI Hard
A. Responsible AI eliminates the need for human oversight, cutting operational headcount significantly
B. Responsible AI is legally mandated in all jurisdictions, so non-compliance guarantees immediate fines
C. Responsible AI reduces long-term costs by preventing brand damage, regulatory penalties, and rework while sustaining customer trust that drives revenue
D. Responsible AI guarantees that models will always achieve higher predictive accuracy than unregulated models

42 An executive claims that responsible AI is 'just a cost center' with no measurable return. Which counterargument most directly reframes responsible AI as a driver of quantifiable business value?

Explain the business case for responsible AI Hard
A. Responsible AI is intangible and its benefits can never be measured in financial terms
B. Trustworthy AI increases user adoption and retention, which can be tracked via engagement metrics and reduced churn
C. Responsible AI should be treated as a marketing slogan rather than an operational practice
D. Responsible AI only matters for companies in regulated industries like finance and healthcare

43 Two competing firms deploy similar AI hiring tools. Firm A invests in bias audits and transparency; Firm B does not. Over three years, which outcome is most consistent with the documented business case for responsible AI?

Explain the business case for responsible AI Hard
A. Firm B outperforms because it ships faster and responsible AI provides no downstream protection
B. Both firms perform identically because responsible AI has no effect on market outcomes
C. Firm A fails because bias audits always render hiring models unusable
D. Firm A incurs higher upfront costs but avoids a discrimination lawsuit and retains partner contracts, while Firm B faces litigation and reputational loss

44 During issue spotting for a loan-approval model, an analyst notices the training data over-represents applicants from affluent postal codes. Which ethical concern is most precisely identified here?

Identify ethical considerations with AI using issue spotting best practices Hard
A. A transparency gap because the model lacks a user-facing explanation
B. Representational bias in the training data that may lead to unfair disparate impact on under-represented groups
C. A privacy violation because postal codes are considered sensitive personal data
D. A security vulnerability enabling model inversion attacks

45 A team uses issue spotting on a generative chatbot. Users can extract snippets of another user's prior conversation through crafted prompts. Which issue category best classifies this finding?

Identify ethical considerations with AI using issue spotting best practices Hard
A. A privacy and data leakage issue where the model exposes information it should keep confidential
B. An accountability issue because no one owns the model's roadmap
C. A fairness issue because outputs differ across user demographics
D. An interpretability issue because the model's weights are opaque

46 When applying issue spotting best practices, why is it recommended to examine the entire AI lifecycle rather than only the deployed model?

Identify ethical considerations with AI using issue spotting best practices Hard
A. Examining the full lifecycle guarantees the model will be entirely free of bias
B. Only the deployment stage can introduce ethical issues, so lifecycle review is a formality
C. Ethical issues can originate at any stage—data collection, labeling, training, deployment, and monitoring—so isolated review misses upstream and downstream harms
D. Lifecycle review is required solely to satisfy documentation audits, not to find real issues

47 An issue-spotting review flags that a medical triage model performs well overall but has a much higher false-negative rate for a minority subgroup. Which best practice most directly addresses this finding?

Identify ethical considerations with AI using issue spotting best practices Hard
A. Remove the subgroup from the dataset to stabilize the aggregate metric
B. Evaluate performance using disaggregated metrics across subgroups rather than relying on a single aggregate score
C. Deploy the model as-is since overall performance exceeds the benchmark
D. Increase the overall accuracy threshold until the aggregate score improves

48 A generative image model reliably produces stereotyped depictions when given occupation prompts. From an issue-spotting standpoint, which framing correctly distinguishes the harm type?

Identify ethical considerations with AI using issue spotting best practices Hard
A. It is an allocative harm because users are denied access to images
B. It is a representational harm that reinforces social stereotypes, distinct from an allocative harm that withholds resources
C. It is a security harm because prompts can be manipulated
D. It is purely a technical accuracy bug with no ethical dimension

49 Google's AI Principles include both objectives to pursue and applications it will not pursue. Which of the following is explicitly listed among the applications Google will not pursue?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Hard
A. Technologies whose principal purpose is to cause or directly facilitate injury to people
B. All research involving large language models
C. Any product that competes with existing open-source tools
D. Any AI system that uses personal data for personalization

50 A key lesson from Google's operationalization of its AI Principles is that principles alone are insufficient. Which mechanism best exemplifies how Google translated principles into practice?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Hard
A. Applying the principles only to research papers, not to deployed products
B. Establishing review processes and governance bodies that assess products against the principles before launch
C. Publishing the principles publicly and relying on employees to self-enforce without any review
D. Delegating all ethical decisions to an external regulator

51 Google emphasizes that its AI Principles are a 'living' framework. What is the primary rationale for treating responsible AI principles as evolving rather than fixed?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Hard
A. Evolving principles allow the company to avoid publishing any concrete commitments
B. AI capabilities, societal expectations, and risks change over time, so guidance must be revisited and refined
C. Changing the principles frequently reduces the need for internal review boards
D. Fixed principles are illegal under most data-protection laws

52 One lesson Google shares is that responsible AI requires cross-functional collaboration. Which scenario best illustrates why a purely engineering-led review is inadequate?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Hard
A. Engineers are unable to write code that meets performance benchmarks without help
B. A model that is technically sound may still cause social or legal harm that requires ethicists, domain experts, and legal input to identify
C. Cross-functional teams are needed only to speed up documentation, not to spot harms
D. Legal teams alone can certify that a model is free of all ethical risk

53 Google's experience shows that operationalizing principles often surfaces tensions between principles. Which example best represents such a tension requiring deliberate trade-off decisions?

Describe how Google developed and put their AI Principles into practice and leverage their lessons learned Hard
A. Improving accuracy always improves privacy simultaneously, so no trade-off exists
B. Improving model accuracy may require more personal data, which can conflict with privacy commitments
C. Transparency and fairness are identical goals and can never conflict
D. Safety and accountability are unrelated and never interact

54 An organization wants to operationalize responsible AI. It has published values but no way to enforce them. According to a maturity-oriented framework, what is the most effective next step?

Adopt a framework for how to operationalize responsible AI in your organization Hard
A. Require every employee to individually interpret the values without shared processes
B. Define concrete governance structures, roles, and review checkpoints that embed the values into workflows
C. Add the values to the company website and consider the work complete
D. Wait for government regulation to dictate exactly what to do

55 A company assigns responsible AI 'ownership' to a single junior analyst with no authority. Why does this undermine operationalization from a framework perspective?

Adopt a framework for how to operationalize responsible AI in your organization Hard
A. A single owner is ideal because it centralizes all decisions in one person
B. Effective operationalization requires accountability backed by executive sponsorship and decision authority across the organization
C. Ownership should always rest with external auditors rather than internal staff
D. Responsible AI does not require any assigned ownership to function

56 When operationalizing responsible AI, why is continuous monitoring after deployment considered essential rather than optional?

Adopt a framework for how to operationalize responsible AI in your organization Hard
A. Continuous monitoring replaces the need for any pre-launch evaluation
B. Model behavior can drift as data and usage change, introducing new harms that pre-launch review cannot anticipate
C. Once a model passes pre-launch review, its behavior is permanently fixed
D. Monitoring is only needed to reduce cloud compute costs

57 An organization's responsible AI framework mandates review only for 'high-risk' systems. Which approach best determines which systems qualify as high-risk?

Adopt a framework for how to operationalize responsible AI in your organization Hard
A. Letting each engineering team self-declare risk with no shared criteria
B. Reviewing only the systems that took the longest to build
C. A risk-tiering assessment based on potential for harm, affected population, and reversibility of decisions
D. Classifying every system that uses cloud infrastructure as high-risk

58 A framework recommends embedding responsible AI 'by design.' Which practice most faithfully reflects this principle?

Adopt a framework for how to operationalize responsible AI in your organization Hard
A. Delegating all responsible AI work to a final QA gate before release
B. Running a single ethics review only after the product ships to customers
C. Incorporating fairness, privacy, and safety requirements into the earliest design and data stages, not as a post-hoc audit
D. Adding a disclaimer to the user interface after launch

59 An organization struggles because its responsible AI policies are ignored under delivery pressure. Which framework intervention most directly addresses this cultural barrier?

Adopt a framework for how to operationalize responsible AI in your organization Hard
A. Aligning incentives and leadership messaging so responsible AI is rewarded rather than seen as a delivery obstacle
B. Increasing the length and detail of the written policy document
C. Moving all reviews to the end of the project to avoid slowing early work
D. Making responsible AI voluntary to reduce friction with deadlines

60 A startup argues it can defer responsible AI until after achieving product-market fit. Which risk most strongly challenges this 'defer it' strategy?

Explain the business case for responsible AI Hard
A. Retrofitting responsibility into a scaled system is far costlier and may require redesigning data pipelines and models already in production
B. Responsible AI can only ever be added before writing the first line of code
C. Product-market fit automatically resolves all ethical concerns
D. Deferring responsible AI has no downside because early-stage products face no scrutiny