Responsible AI Field Manual for Monitoring & Evaluation
EvalCommunity Academy Tutorial
How to Use the Responsible AI Field Manual for Monitoring & Evaluation
Learn how to classify AI risk, protect evidence quality, support ethical practice, and use AI responsibly across the Monitoring and Evaluation cycle.
10 min read
Beginner-friendly
For M&E professionals
What this tutorial covers
The Responsible AI Field Manual helps evaluators, consultants, programme teams, and organisations decide when AI is appropriate, what safeguards are needed, and how to protect professional judgement.
- Classify AI use by risk level.
- Check whether sensitive data can be used safely.
- Protect evidence quality and human accountability.
- Document how AI was used in an evaluation project.
- Use templates, tools, and checklists for responsible practice.
Step 1: Start with the Principles
Begin by reviewing the Principles section. This explains the core commitments behind responsible AI use in M&E: human accountability, evidence connection, proportional safeguards, data care, context awareness, and transparent records.
The key message is simple: AI can support M&E work, but professional judgement remains with people.
Step 2: Use the Risk Ladder
The Risk Ladder helps you classify AI use from low-risk support to uses that may require formal approval or should be avoided.
Level 1: Routine support
Proofreading, formatting, plain-language editing, or agenda drafting.
Level 2: Evidence organisation
Summarising public documents, structuring notes, or creating synthesis tables.
Level 3: Analytical assistance
Suggesting codes, themes, comparisons, categories, or analytical patterns.
Level 4: Evaluative interpretation
Work that may influence findings, recommendations, causal claims, or performance judgements.
Level 5: Normally inappropriate
Replacing professional judgement, using sensitive data without approval, or simulating stakeholder voices.
Step 3: Review AI use across the M&E cycle
Use the M&E Cycle section to understand where AI might appear during scoping, design, collection, data management, analysis, reporting, and learning.
At each stage, ask what AI is helping with, what evidence is being used, what could go wrong, and who will review the final output.
Step 4: Apply the Quality Assurance checklist
Before any AI-assisted work influences decisions, check whether the output is accurate, evidence-based, and professionally defensible.
- Is the output supported by the source material?
- Has anything important been omitted or overstated?
- Are minority or dissenting views still visible?
- Are causal claims justified by the evidence?
- Could the evaluator defend the conclusion without relying on AI wording?
Step 5: Check ethics, equity, and data sensitivity
Use the Ethics & Equity section before using AI with interview transcripts, personal data, vulnerable groups, confidential information, politically sensitive material, or community narratives.
AI should not flatten lived experience, erase local meaning, or expose information that should remain protected.
Step 6: Set governance rules
The Governance section helps teams agree on which AI uses are allowed, which require approval, what data must never be entered into AI tools, who reviews outputs, and how AI use will be disclosed.
Good governance questions
- Which AI uses are allowed, conditional, or prohibited?
- What data must never be entered into AI tools?
- Who approves higher-risk uses?
- How will AI use be disclosed to clients, funders, or participants?
Step 7: Use the Field Manual tools
The Field Manual includes practical tools that teams can use before, during, and after AI-supported work.
1. AI Use Register
Record the AI tool used, task type, data type, risk level, review method, disclosure decision, and responsible person.
2. Evidence-to-Claim Map
Connect each major finding, conclusion, or recommendation to the evidence that supports it.
3. Sensitive Data Decision Tool
Check whether data is public, internal, confidential, personal, politically sensitive, or protected by consent conditions.
4. Professional Judgement Check
Confirm that the evaluator can explain the finding, uncertainty, alternative explanations, and limits without relying only on AI wording.
5. Equity and Voice Review
Check whether smaller groups, dissenting views, local categories, and marginalised perspectives have been preserved.
6. Disclosure Builder
Prepare clear language explaining what AI was used for, what it was not used for, and how human review was maintained.
Step 8: Try the Risk Checker
The Risk Checker helps classify a planned AI use by task type, data sensitivity, decision influence, and review route. Use it before beginning AI-supported work, especially when the work involves analysis, findings, recommendations, or sensitive information.
Step 9: Use the Templates
The Templates section gives teams ready-to-use language and documentation formats for registers, disclosure notes, review records, and quality assurance checks.
Step 10: Build an operating rhythm
Responsible AI should become part of how teams plan, review, and learn. Agree AI rules at project start, check risk before use, review evidence during analysis, validate findings before delivery, and document lessons after delivery.
Start using the Responsible AI Field Manual
Apply responsible AI safeguards to your next Monitoring and Evaluation task. Use the manual to classify risk, protect evidence quality, and maintain professional accountability.
Continue Learning with EvalCommunity Academy
Take the AI in Monitoring & Evaluation course
Build practical skills for using AI responsibly in M&E workflows, from evidence review and reporting to governance, quality assurance, and ethical safeguards.
