Wayground logo

Free Printable Worksheets

Font size

S
M
L
XL
Worksheets

AI nondiscrimination worksheet (grade 14)

Total questions: 70

Worksheet time: 35mins

Name
Class
Date
1.

Bias persistence: Why does removing protected attributes from training data often fail to prevent discrimination?

a)

Protected attributes are legally required

b)

Correlated variables can still encode protected characteristics

c)

Bias only occurs at inference time

d)

AI models cannot process sensitive data

2.

Data drift and discrimination: Why is keeping training data up to date important for nondiscrimination compliance?

a)

Older data violates copyright

b)

Societal and demographic shifts can change how risk factors impact groups over time

c)

Regulators require monthly retraining

d)

AI models decay automatically

3.

Privacy-fairness tension: Why do privacy laws complicate bias testing in AI?

a)

Bias testing is prohibited

b)

Sensitive attributes may be restricted, limiting direct fairness measurement

c)

Fairness overrides privacy

d)

Only anonymised data may be used

4.

Generative AI complexity: What makes generative AI systems harder to regulate for nondiscrimination?

a)

They cannot be audited

b)

Bias can emerge at multiple stages across prompts, intermediate steps and outputs

c)

They do not use training data

d)

They only operate in unregulated sectors

5.

Subjective fairness: Why do technical teams often disagree on whether an AI outcome is discriminatory?

a)

Laws define discrimination numerically

b)

Fairness definitions depend on normative and contextual judgments

c)

Discrimination only applies to intent

d)

AI accuracy resolves fairness

6.

Section 1557 trigger: Under HHS OCR Section 1557, when does an AI tool create compliance risk?

a)

When accuracy falls below benchmark

b)

When it produces discriminatory impacts in covered health care settings

c)

When it is experimental

d)

When it processes non-clinical data

7.

Proactive duty: Section 1557 is notable because it requires covered entities to:

a)

Respond only after complaints

b)

Proactively identify and correct discriminatory AI impacts

c)

Obtain pre-approval for AI

d)

Publish source code

8.

21st Century Cures Act relevance: Why is the 21st Century Cures Act relevant to AI nondiscrimination?

a)

It regulates model architectures

b)

It promotes transparency and access to health data to reduce inequities

c)

It bans AI diagnostics

d)

It replaces HIPAA

9.

Health AI misconception: Which belief creates the highest compliance risk in health AI?

a)

“AI assists clinicians”

b)

“Clinical oversight eliminates discrimination risk”

c)

“AI accuracy is high”

d)

“AI improves efficiency”

10.

Administrative AI risk: Which non-clinical AI use in health care still raises nondiscrimination concerns?

a)

Billing prioritisation

b)

Appointment scheduling and patient triage

c)

Inventory forecasting

d)

Facility maintenance

11.

Lawful vs unlawful discrimination: What distinguishes lawful risk-based pricing from unlawful discrimination?

a)

Whether AI is used

b)

Whether differentiation is actuarially justified and not based on protected traits

c)

Whether customers consent

d)

Whether the model is explainable

12.

AI amplification risk: Why does AI heighten discrimination risk in insurance underwriting?

a)

AI ignores actuarial data

b)

AI may identify proxies for protected characteristics at scale

c)

AI eliminates regulation

d)

AI replaces state oversight

13.

NAIC Model Law role: The NAIC Model Law (2020) primarily serves to:

a)

Ban AI in insurance

b)

Guide states on regulating AI to prevent unethical or unlawful discrimination

c)

Replace actuarial standards

d)

Federalise insurance law

14.

Claims processing bias: Which AI insurance function is often overlooked but high risk?

a)

Marketing personalisation

b)

Claims fraud detection

c)

Website chatbots

d)

Product naming

15.

New York insurance guidance: New York’s AI guidance for insurers is notable because it:

a)

Applies only to pricing

b)

Applies to all authorised insurers using AI

c)

Is voluntary

d)

Replaces NAIC guidance

16.

Core legal exposure: Why does AI screening trigger heightened legal scrutiny in employment?

a)

Employment law bans automation

b)

Hiring decisions affect access to work, a protected legal interest

c)

AI lacks transparency

d)

Models are experimental

17.

EEOC’s key principle: The EEOC’s position on AI in hiring can be summarised as:

a)

AI decisions are presumptively lawful

b)

AI must meet the same nondiscrimination standards as human decisions

c)

AI requires new employment law

d)

AI is advisory only

18.

Disparate impact example: Which hiring practice most clearly illustrates disparate impact risk?

a)

Rejecting all candidates without degrees

b)

Screening out candidates based on resume gaps correlated with caregiving responsibilities

c)

Requiring job-specific certifications

d)

Manual interview scoring

19.

NYC audit requirement: NYC’s automated employment decision law requires:

a)

Model disclosure

b)

Independent bias audits and candidate notification

c)

Federal approval

d)

Consent from all applicants

20.

Oversight fallacy: Which assumption most often leads to EEOC violations?

a)

“We test accuracy”

b)

“The AI only recommends”

c)

“We audit for bias”

d)

“We retrain regularly”

21.

Foundational statute: Which law underpins nondiscrimination expectations in AI credit decisions?

a)

GDPR

b)

FCRA

c)

HIPAA

d)

FHA

22.

AI-specific gap: The absence of AI-specific credit law means:

a)

AI credit systems are unregulated

b)

Existing financial laws must be applied to AI contexts

c)

States regulate exclusively

d)

Accuracy is the only requirement

23.

CFPB signal: The CFPB’s 2021 RFI suggests the agency is:

a)

Approving AI credit scoring

b)

Monitoring AI practices ahead of guidance or enforcement

c)

Delegating oversight

d)

Ending FCRA enforcement

24.

Explainability importance: Why is explainability critical in AI credit systems?

a)

For public trust

b)

Because consumers have rights to understand adverse decisions

c)

To improve model accuracy

d)

To reduce compute cost

25.

Proxy risk in lending: Which variable poses the highest proxy discrimination risk?

a)

Loan amount

b)

Geographic indicators tied to historical segregation

c)

Interest rate

d)

Loan term

26.

Foundational housing law: Which statute governs AI-based housing decisions?

a)

FCRA

b)

FHA

c)

EEOC Act

d)

NAIC Model Law

27.

Ranking misconception: Why is “we only rank applicants” an invalid defence?

a)

Ranking is illegal

b)

Rankings still influence housing access outcomes

c)

Rankings are always biased

d)

Rankings are unregulated

28.

HUD guidance focus: HUD’s 2020 guidance emphasises that:

a)

AI is prohibited in housing

b)

Automated systems must adhere to existing fair housing obligations

c)

Only lenders are covered

d)

Consent removes liability

29.

Review necessity: Why must housing AI systems be regularly reviewed?

a)

Models age quickly

b)

Housing market dynamics and demographic effects change

c)

HUD mandates monthly audits

d)

AI accuracy degrades

30.

Tenant screening risk: Which tenant screening signal is most likely to raise FHA concerns?

a)

Prior eviction history

b)

Criminal background proxies tied to protected groups

c)

Income verification

d)

Lease duration preference

31.

Audit function: Why are bias audits central to nondiscrimination governance?

a)

They replace regulators

b)

They detect, document and enable remediation of discriminatory outcomes

c)

They guarantee compliance

d)

They eliminate human oversight

32.

Audit timing: Which audit approach is most defensible?

a)

One-time pre-deployment testing

b)

Continuous or periodic audits across the system lifecycle

c)

Only complaint-driven audits

d)

Vendor-only audits

33.

Vendor risk: Why do third-party AI tools increase nondiscrimination risk?

a)

Vendors assume liability

b)

Organisations remain responsible for outcomes even when tools are procured

c)

Vendors hide bias

d)

Vendors cannot be audited

34.

Human oversight myth: Why is “human in the loop” not sufficient by itself?

a)

Humans always agree with AI

b)

Humans may over‑rely on AI recommendations

c)

Laws ban automation

d)

Oversight eliminates bias

35.

Training effectiveness: Why is staff training part of nondiscrimination compliance?

a)

It reduces compute cost

b)

It helps decision‑makers understand AI limitations and bias risks

c)

It replaces audits

d)

It improves accuracy

36.

Best governance response: An AI hiring tool shows lower pass rates for a protected group. What is the most defensible response?

a)

Disable the model

b)

Investigate data, features, thresholds and decision context

c)

Increase accuracy

d)

Ignore if intent was neutral

37.

Cross‑sector pattern: Across employment, credit and housing, nondiscrimination law focuses most on:

a)

Intent

b)

Impact

c)

Model size

d)

Training time

38.

Privacy‑safe bias testing: Which approach best balances privacy and fairness?

a)

Avoid testing

b)

Use protected attributes under strict governance solely for bias detection

c)

Publish personal data

d)

Rely on synthetic proxies only

39.

Generative AI hiring risk: Why does generative AI introduce new hiring risks?

a)

It replaces HR

b)

It may embed biased language patterns into screening or evaluation outputs

c)

It lacks accuracy

d)

It cannot be audited

40.

Insurance claims synthesis: An AI flags claims from a specific demographic more often for fraud review. What is the primary legal risk?

a)

Accuracy loss

b)

Disparate treatment in claims processing

c)

IP infringement

d)

Data breach

41.

Credit decision escalation: When does AI credit scoring most likely trigger regulatory scrutiny?

a)

When accuracy is low

b)

When adverse decisions cannot be explained meaningfully

c)

When models are proprietary

d)

When models are fast

42.

Housing ranking scenario: An AI ranks applicants but landlords decide manually. Liability risk exists because:

a)

AI made no decision

b)

Rankings influence outcomes and access

c)

FHA does not apply

d)

Consent was given

43.

Insurance lawful discrimination boundary: Which factor most strongly legitimises differential pricing?

a)

Business convenience

b)

Actuarial evidence and regulatory acceptance

c)

Model explainability

d)

Customer consent

44.

Governance red flag: Which is the strongest red flag for nondiscrimination risk?

a)

High accuracy

b)

No bias monitoring after deployment

c)

Vendor certification

d)

Human oversight

45.

Multi‑law exposure: Why do AI systems often trigger multiple nondiscrimination regimes simultaneously?

a)

AI is global

b)

Single decisions affect multiple legally protected interests

c)

Laws overlap accidentally

d)

Regulators coordinate

46.

Best documentation practice: Which documentation best supports nondiscrimination defence?

a)

Model weights

b)

Bias audit results and remediation actions

c)

Marketing materials

d)

Source code

47.

Insurance proxy trap: Why are lifestyle variables risky in insurance AI?

a)

They are inaccurate

b)

They can correlate with protected traits

c)

They reduce performance

d)

They are illegal

48.

Employment AI risk escalation: Which change most increases legal risk?

a)

New UI

b)

Expanding AI from resume screening to promotion decisions

c)

Retraining model

d)

Adding human review

49.

Housing compliance defence: Which argument is weakest in defending an AI housing system?

a)

“We regularly test for disparate impact”

b)

“We rely on historical data”

c)

“We corrected identified issues”

d)

“We document decisions”

50.

Final exam takeaway: Which statement best captures AIGP’s nondiscrimination stance on AI?

a)

AI is neutral if data is neutral

b)

Accuracy ensures fairness

c)

Nondiscrimination compliance requires continuous, context‑aware governance

d)

Laws will adapt later

51.

Why AI hiring tools trigger nondiscrimination scrutiny: AI tools used in hiring are closely scrutinised under nondiscrimination law primarily because they:

a)

Replace all human judgement

b)

Influence access to employment opportunities, which is a protected legal interest

c)

Always use sensitive personal data

d)

Are regulated by AI-specific employment statutes

52.

EEOC’s core position on AI in hiring: The EEOC’s 2021 guidance makes clear that AI in hiring:

a)

Is exempt from federal nondiscrimination laws

b)

Is permitted only if accuracy exceeds a set threshold

c)

Must comply with existing federal nondiscrimination laws in the same way as human decision-making

d)

Requires prior EEOC approval

53.

Disparate impact risk: Which scenario best illustrates a disparate impact risk in AI hiring?

a)

An AI tool rejects all candidates without degrees

b)

An AI tool disproportionately screens out candidates from a protected group, even without explicit use of protected attributes

c)

An AI tool is trained on synthetic data

d)

An AI tool explains its recommendations

54.

NYC Local Law on AI hiring: New York City’s local law on automated employment decision tools is notable because it:

a)

Bans AI from hiring entirely

b)

Requires public disclosure of model source code

c)

Mandates a specific bias audit before use in employment decisions affecting NYC workers

d)

Applies only to federal agencies

55.

Human oversight misconception: Which statement best reflects a common compliance misconception in AI hiring?

a)

Human review eliminates all legal risk

b)

If a human can override AI, discrimination law does not apply

c)

AI outputs must still be assessed for discriminatory impact even with human oversight

d)

Oversight is required only for fully automated decisions

56.

Foundational law for credit nondiscrimination: In the U.S., the foundational law governing nondiscrimination in credit and lending is:

a)

The GDPR

b)

The Fair Credit Reporting Act (FCRA)

c)

An AI-specific federal statute

d)

The Equal Pay Act

57.

Lack of AI-specific statute: The absence of AI-specific credit legislation primarily means that:

a)

AI use in credit is unregulated

b)

Existing financial services laws must be interpreted and applied to AI systems

c)

Only state laws apply

d)

Credit AI systems cannot be audited

58.

CFPB’s 2021 action: The CFPB’s 2021 request for information on AI in credit decision-making signals that the agency:

a)

Approved all AI credit scoring

b)

Sought to understand risks, transparency gaps and consumer harms before issuing formal rules

c)

Delegated oversight to private auditors

d)

Focused only on cybersecurity

59.

Explainability requirement: Why is explainability particularly important in AI-based credit decisions?

a)

Because models must be open source

b)

Because consumers have rights to understand adverse credit decisions

c)

Because explainability improves accuracy

d)

Because regulators require model weights

60.

Proxy discrimination risk: Which feature presents the greatest proxy discrimination risk in AI credit models?

a)

Interest rate caps

b)

Variables strongly correlated with protected characteristics, such as postcode or employment history patterns

c)

Loan amount requested

d)

Repayment schedule

61.

Foundational housing law: The primary U.S. law governing housing discrimination remains the:

a)

Civil Rights Act (Title VII)

b)

Fair Housing Act (1968)

c)

Consumer Credit Protection Act

d)

AI Accountability Act

62.

AI use in housing: AI systems used for tenant screening or housing decisions must:

a)

Be certified by HUD

b)

Demonstrate compliance with fair housing rules even when ranking or scoring applicants

c)

Avoid using any personal data

d)

Be fully automated

63.

HUD’s 2020 guidance focus: HUD’s 2020 guidance primarily addressed:

a)

Energy efficiency in housing

b)

Automated decision-making in rental and mortgage contexts

c)

AI explainability standards

d)

Smart city development

64.

Compliance misconception in housing AI: Which belief is most likely to lead to housing law violations?

a)

“We use AI only as a recommendation tool”

b)

“The Fair Housing Act does not apply if decisions are data-driven”

c)

“Regular reviews reduce risk”

d)

“AI outputs should be monitored over time”

65.

Ongoing review importance: Why does the module emphasise regular review of housing AI systems?

a)

Models degrade technically

b)

Housing markets and populations change, altering impact across protected groups

c)

Regulators require weekly audits

d)

AI outputs are legally binding

66.

Across hiring, credit and housing, nondiscrimination law focuses primarily on

a)

Model transparency

b)

Outcomes and effects on protected groups

c)

Algorithmic novelty

d)

Data volume

67.

Why are bias audits particularly important in these regulated domains?

a)

They replace regulators

b)

They help detect, document and remediate discriminatory patterns before enforcement

c)

They eliminate the need for human oversight

d)

They guarantee compliance

68.

Why do generative AI systems increase nondiscrimination risk?

a)

They are always opaque

b)

They involve multi-step pipelines where bias can be introduced at multiple stages

c)

They replace all traditional models

d)

They cannot be tested

69.

Why do privacy laws complicate nondiscrimination compliance in AI?

a)

Privacy laws ban bias testing

b)

Restrictions on collecting sensitive attributes can limit the ability to detect and measure bias

c)

Privacy laws override employment law

d)

Privacy laws require anonymisation only

70.

Which governance approach best aligns with the module’s guidance across all three domains?

a)

Avoid AI entirely

b)

Rely on vendor assurances

c)

Combine legal review, bias audits, human oversight, and continuous monitoring

d)

Focus only on model accuracy