QUIZ: Preparing Submissions for AI-IRB Approval

Assessment settings

• Passing grade required: Yes
• Passing grade: 80%

Pass message

Excellent Work!
Your answers reflect a solid understanding of the documentation, ethical safeguards, and responsiveness required for AI-IRB approvals. Keep these insights in mind to ensure your AI research proposals align with standards protecting both participants and society.

Fail message

Thank You for Completing the Quiz.
Your score suggests revisiting the fundamentals—particularly the specifics of informed consent, data management, and addressing algorithmic bias. Review the lesson on how to craft a clear, thorough research protocol, then retake the quiz once you’ve refreshed these concepts.

Questions and answer key

1. What primary role does the AI-IRB play when assessing AI research proposals? (Select all that apply.)

Type: multiple

• A. Ensuring compliance with local data protection laws and regulations
• B. Verifying that the project uses only open-source AI tools
• C. Evaluating participant safety and ethical considerations unique to AI systems
• D. Reviewing the project’s marketing strategy for software commercialization

Explanation

Correct Answers: A, C
Explanation:

• A. The IRB checks adherence to data privacy/security standards (e.g., GDPR, HIPAA).

• C. AI-IRBs consider AI-specific risks like algorithmic bias and data misuse to protect participants.

• B is not required; whether the AI solution is open-source or proprietary doesn’t inherently factor into ethical approval.

• D. The IRB’s scope is ethical oversight, not market viability.

2. Which essential document should researchers prepare to clarify how they collect, process, and protect data in AI studies? (Select all that apply.)

Type: multiple

• A. A generic letter stating that no personal data is ever used
• B. A data management plan detailing encryption, access controls, and anonymization methods
• C. A code repository release form requiring all AI source code be made public
• D. An in-depth risk assessment describing potential data breaches and mitigation strategies

Explanation

Correct Answers: B, D
Explanation:

• B. Researchers must outline how they’ll manage data (including privacy safeguards).

• D. A thorough risk analysis demonstrates awareness of vulnerabilities and how to address them.

• A is too vague; many AI projects do use personal or sensitive data.

• C is typically not mandatory for IRB approval (open-sourcing code is a separate decision).

3. When developing the research protocol for AI projects, which elements are most crucial to include? (Select all that apply.)

Type: multiple

• A. Clear aims and hypotheses describing the study’s objectives
• B. A reference to unverified internet articles without peer-reviewed support
• C. Detailed methodology explaining which AI algorithms and datasets are used
• D. A discussion of how participants’ autonomy and consent rights will be upheld

Explanation

Correct Answers: A, C, D
Explanation:

• A. A well-defined purpose helps the IRB grasp the rationale behind the AI research.

• C. The IRB wants clarity on model choice, dataset sources, and data usage.

• D. Safeguarding human subjects includes ensuring a transparent consent process.

• B. Over-reliance on unverified information weakens credibility and may raise concerns about data quality.

4. Which considerations commonly appear in an informed consent form for AI research? (Select all that apply.)

Type: multiple

• A. Explanation of how participants’ data will be collected and analyzed by AI algorithms
• B. Guarantee that the system will never produce biased outcomes, under any circumstances
• C. Potential risks, including data privacy concerns and the possibility of algorithmic misinterpretation
• D. The option for participants to withdraw from the study without penalty at any stage

Explanation

Correct Answers: A, C, D
Explanation:

• A. Participants deserve clarity about AI’s role in data analysis.

• C. They must be informed of possible privacy or bias risks.

• D. IRB ethics requires voluntariness, ensuring participants can exit.

• B is unrealistic; no AI project can promise zero bias in all scenarios.

5. Common pitfalls that may cause delays or rejections in AI-IRB submissions include: (Select all that apply.)

Type: multiple

• A. Vague explanations of how AI techniques will be used in the research
• B. Thoroughly addressing risk mitigation strategies for data privacy
• C. Overlooking potential algorithmic biases that could affect participant groups
• D. Inadequate disclosure regarding participants’ ability to withdraw from the study

Explanation

Correct Answers: A, C, D
Explanation:

• A. If it’s unclear how AI is applied, the IRB can’t properly evaluate ethical implications.

• C. Ignoring bias signals potential harm to participants or certain demographics.

• D. Participants must be told they can withdraw at any time; omitting this detail is a red flag.

• B is actually good practice and not a pitfall.

6. In addressing bias within AI algorithms for research, the IRB often expects researchers to: (Select all that apply.)

Type: multiple

• A. Demonstrate methods for identifying or reducing potential bias in training data
• B. Provide references to prior works that trivialize algorithmic fairness concerns
• C. Validate their models using diverse test sets to evaluate equitable performance
• D. Assume that any large dataset is inherently unbiased, no additional checks needed

Explanation

Correct Answers: A, C
Explanation:

• A. Proactive steps to spot and mitigate bias show ethical diligence.

• C. Using broad, representative datasets and analyzing outcomes across subpopulations is crucial.

• B indicates a dismissive stance toward fairness, undermining ethical rigor.

• D. Large data can still contain historical or systemic biases; always needs scrutiny.

7. During the IRB review process, which actions by researchers can improve the likelihood of approval? (Select all that apply.)

Type: multiple

• A. Preemptively addressing common ethical questions, such as privacy measures and potential participant risks
• B. Using overly complex jargon in the protocol to highlight technical expertise
• C. Responding promptly to IRB feedback and revising documents as requested
• D. Providing incomplete documentation in hopes the IRB overlooks minor gaps

Explanation

Correct Answers: A, C
Explanation:

• A. Anticipating typical IRB concerns (data handling, participant safety) reassures reviewers.

• C. Timely responsiveness signals a cooperative and organized approach to ethics.

• B may confuse IRB members, hindering clarity and raising further questions.

• D is unwise; incomplete info stalls reviews or leads to rejections.

8. To strengthen the rationale behind an AI study, researchers can include: (Select all that apply.)

Type: multiple

• A. Pilot data or prior research illustrating the need for the AI solution
• B. Claims of “unlimited benefits” without tangible evidence
• C. A literature review discussing existing standards/guidelines for ethical AI research
• D. The project’s estimated market value to prove potential profitability

Explanation

Correct Answers: A, C
Explanation:

• A. Early data or prior findings support the study’s feasibility and ethical preparedness.

• C. Contextualizing the work with established best practices shows diligence in ethical AI use.

• B. Exaggerated claims lacking support undermine credibility.

• D. Market value isn’t generally an IRB concern; they focus on ethical, not commercial, viability.

9. If researchers need clarification on IRB requirements mid-process, recommended approaches include: (Select all that apply.)

Type: multiple

• A. Engaging in open discussions with the IRB office or designated contacts
• B. Assuming any ambiguous requirement is optional, proceeding without changes
• C. Reviewing official institutional guidelines or seeking mentorship from experienced peers
• D. Submitting an entirely different proposal to skip potential IRB scrutiny

Explanation

Correct Answers: A, C
Explanation:

• A. IRB administrators can clarify submission specifics or policy nuances.

• C. Institutional documents and experienced colleagues can offer valuable guidance.

• B. Skipping unclear steps may cause misunderstandings or rejections later.

• D. Replacing the entire proposal doesn’t address actual IRB questions or concerns.

10. Upon receiving IRB approval, researchers should: (Select all that apply.)

Type: multiple

• A. Immediately modify protocols mid-study without notifying the IRB
• B. Adhere to all approved procedures and conditions throughout the study
• C. Continue monitoring AI methods for new biases or ethical issues
• D. View IRB oversight as concluded, with no further reporting requirements

Explanation

Correct Answers: B, C
Explanation:

• B. Maintaining the IRB-approved protocols prevents unapproved changes that can invalidate approval.

• C. Ongoing vigilance is needed because AI systems may evolve or reveal unanticipated biases over time.

• A. Changing procedures unilaterally undermines the IRB’s review; amendments must be submitted if significant.

• D. Researchers typically file progress or final reports, especially if there are protocol changes or incidents.