AI Prompt for Ethical Risk Assessment Prompt for Research Proposals

Objective: Create a reusable AI prompt template that systematically guides ethical and risk assessment of research proposals by examining dual-use risks, privacy implications, fairness issues, and unintended consequences while citing relevant guidelines and historical precedents.

How the Base Prompt Was Formed

The base prompt was formed through these deliberate design steps:

Identify the core analytical sequence required for thorough ethical review so the AI always follows the same logical order.
Isolate fixed instructional language that remains constant across uses and convert those passages into reusable snippets.
Extract elements that must change with each new proposal or audience and convert them into variables.
Embed explicit requirements to reference established frameworks and real historical cases so the output stays grounded and evidence-based.
Add structural markers that force step-by-step reasoning and clear separation of each risk category.

Note

Keeping the analytical order fixed (dual-use → privacy → fairness → unintended consequences) prevents the AI from skipping categories or mixing concerns.

Base Prompt Template

Prompt Template

{{snippet:brand_voice}}

You are an expert ethics reviewer. Conduct a structured ethical and risk assessment of the following research proposal.

Proposal title: {{proposal_title}}
Proposal summary: {{proposal_summary}}
Target audience: {{target_audience}}
Assessment tone: {{tone}}

Proceed step by step through the four required areas:

1. Dual-use risks
2. Privacy implications
3. Fairness issues
4. Unintended consequences

For each area:
- Describe potential risks in concrete terms.
- Cite at least one relevant guideline or framework (for example Belmont Report, UNESCO Recommendation on the Ethics of AI, or institutional dual-use policies).
- Reference one historical precedent that illustrates the risk.
- Suggest practical mitigation measures.

{{snippet:citation_and_precedent_rules}}

End with a concise overall risk rating and a prioritized list of recommended actions.

Snippets and Variables

Element	Type	Example Value
{{snippet:brand_voice}}	snippet	Maintain a clear, balanced, and professional tone suitable for institutional ethics review committees.
{{proposal_title}}	variable	Genomic Surveillance of Emerging Pathogens
{{proposal_summary}}	variable	A project that develops AI models to predict pathogen evolution from open genomic data and clinical metadata.
{{target_audience}}	variable	University research ethics board
{{tone}}	variable	formal and precise
{{snippet:citation_and_precedent_rules}}	snippet	Always name the specific guideline or framework and the historical case. Prefer publicly available sources such as Belmont Report principles, dual-use research of concern criteria, or well-documented incidents such as the 2011 H5N1 transmission studies.

Tip

Place citation and historical-precedent requirements inside a snippet so they appear consistently without rewriting them for every new proposal.

Filled Base Prompt

Prompt Example

Maintain a clear, balanced, and professional tone suitable for institutional ethics review committees.

You are an expert ethics reviewer. Conduct a structured ethical and risk assessment of the following research proposal.

Proposal title: Genomic Surveillance of Emerging Pathogens
Proposal summary: A project that develops AI models to predict pathogen evolution from open genomic data and clinical metadata.
Target audience: University research ethics board
Assessment tone: formal and precise

Proceed step by step through the four required areas:

1. Dual-use risks
2. Privacy implications
3. Fairness issues
4. Unintended consequences

For each area:
- Describe potential risks in concrete terms.
- Cite at least one relevant guideline or framework (for example Belmont Report, UNESCO Recommendation on the Ethics of AI, or institutional dual-use policies).
- Reference one historical precedent that illustrates the risk.
- Suggest practical mitigation measures.

Always name the specific guideline or framework and the historical case. Prefer publicly available sources such as Belmont Report principles, dual-use research of concern criteria, or well-documented incidents such as the 2011 H5N1 transmission studies.

End with a concise overall risk rating and a prioritized list of recommended actions.

Extended Example Values

{{snippet:brand_voice}}
- Maintain a clear, balanced, and professional tone suitable for institutional ethics review committees.
- Adopt a cautious yet constructive voice that highlights both risks and feasible safeguards.
- Use language appropriate for funding agency reviewers who value transparency and accountability.
{{proposal_title}}
- Genomic Surveillance of Emerging Pathogens
- Large-Scale Social Media Sentiment Analysis for Public Health
- Autonomous Drone Mapping of Protected Habitats
{{proposal_summary}}
- A project that develops AI models to predict pathogen evolution from open genomic data and clinical metadata.
- Collection and analysis of public social media posts to detect early signals of disease outbreaks.
- Deployment of autonomous aerial systems to generate high-resolution ecological maps in conservation zones.
{{target_audience}}
- University research ethics board
- National research funding agency
- Institutional biosafety committee
{{tone}}
- formal and precise
- measured and advisory
- rigorous yet accessible
{{snippet:citation_and_precedent_rules}}
- Always name the specific guideline or framework and the historical case. Prefer publicly available sources such as Belmont Report principles, dual-use research of concern criteria, or well-documented incidents such as the 2011 H5N1 transmission studies.
- Reference at least one international standard and one real-world case for every risk category.
- Prioritize guidelines that are openly accessible and historically significant cases that produced lasting policy changes.