AI as Assistant, Not Replacement
AI does the initial legwork. It doesn't replace critical thinking, source verification, or community validation.
What is MCP (Model Context Protocol)?
MCP structures your prompt so AI gives focused outputs, not generic ones: context, objectives, format, and evidence expectations.
MCP formalizes this into four structured blocks, covered next.
Why Use AI for Problem Analysis?
⚡
Speed
A preliminary Problem Tree in 30-40 minutes instead of a full day.
🌐
Scope
A wide survey of perspectives and sources you might miss otherwise.
🗂️
Structure
Outputs pre-formatted as Problem Tree bullets, indicator tables, or question lists.
📚
Sources
Credible reports from institutions that don't rank highly in a search engine.
❓
Questions
2-3 open-ended validation questions for each assumption in your tree.
What AI Can and Cannot Do
✅ AI Excels At:
- Summarizing research and patterns across sources
- Suggesting cause-effect relationships from literature
- Formatting outputs into trees, tables, or lists
- Drafting validation questions from your assumptions
❌ AI Cannot:
- Replace community voices or lived experience
- Guarantee current accuracy: it may cite outdated information
- Understand local cultural nuance
- Replace your critical thinking: every output needs review
The 70/30 Rule
Spend 70% of effort on the initial AI-assisted scan, 30% on verification and stakeholder prep.
The Research Workflow
- Define your problem scope (15 minutes)
- Customize the MCP prompt (10 minutes)
- Execute AI research (5 minutes)
- Quality verification (20 minutes)
- Build the Problem Tree (25 minutes)
- Prepare validation questions (15 minutes)
Total time: ~90 minutes for a preliminary Problem Tree ready for validation.
Critical Success Factors
- Be specific: generic input produces generic output
- Always verify sources: spot-check 3-5 key citations
- Tag everything (E) or (A) unless you've verified it
- Prepare to be surprised: the goal is to learn what AI missed
Ethical Considerations
- Data privacy: keep prompts general, no identifiable community details
- Bias awareness: be critical when researching marginalized communities
- Community partnership: AI research is preliminary, stakeholder input is definitive