Module 05 · Tier 4 — Intermediate Prompting

AI for Strategic Research

Go beyond search. Use AI to synthesize, compare, and generate insights from complex information.

12 min Tier 4 — Intermediate Prompting

Why This Matters

Strategic research isn't finding information — it's transforming information into insight. AI is exceptionally good at the transformation step: taking a body of information you've gathered and helping you find patterns, surface tensions, generate frameworks, and identify what's missing. The limitation is the same as always: it can't replace your access to the right sources, and it can't verify its own factual claims. This module is about maximizing the transformation value while managing the verification risk.

The Concept

The Research Stack: Where AI Fits

Think of strategic research as a stack of three layers:

Source gathering: Finding the right information (primary sources, databases, interviews, documents). AI is unreliable here for specific facts — use it to generate search queries, identify what sources you need, and map the information landscape.
Synthesis: Extracting meaning from gathered information. AI is excellent here — especially when you provide the source material. Pattern identification, comparison, framework generation, gap analysis.
Implication: What does this mean for your specific situation? This requires your judgment, organizational context, and domain expertise. AI can suggest hypotheses; only you can evaluate them against reality.

The Synthesis Toolkit

Comparative analysis: "Given these three [reports/positions/options], compare them across these dimensions: [list]. Then identify the most significant point of disagreement and the most important question they collectively leave unanswered."

Pattern extraction: "I'm going to paste five [customer interviews/analyst reports/case studies]. After reading all of them, identify: the three themes that appear most consistently, and two patterns that appear in some but are notably absent in others."

Gap mapping: "Based on this research, what important questions remain unanswered? What would I need to know to be more confident in a decision based on this information?"

Framework generation: "Given this information, what are the most useful ways to organize or categorize it? Generate three different frameworks for thinking about this problem and explain what each one reveals."

Steelmanning: "Here is [my position/proposed decision]. Play devil's advocate: what is the strongest possible case against this? What evidence would most concern someone who disagreed?"

Managing Hallucination Risk in Research

In research contexts, hallucination risk is highest for: statistics, citations, specific claims about real organizations or people, and anything AI generates without a document you provided.

The grounding rule: For any specific claim that will inform a decision — especially numerical claims — ask: "Is this from a document I provided, or from training data?" If the latter, verify before using.

The citation test: Ask AI to cite its sources for any factual claim. If it can't produce a specific, verifiable citation, treat the claim as a hypothesis to verify, not a fact to use.

Synthesis vs. search: the same research question

The weak approach: "What are the main trends in [industry]?" → AI generates a generic trends list from training data. Plausible, but not grounded in current sources, not specific to your question, and not actionable.

The synthesis approach: Gather 5-7 recent sources (analyst reports, news articles, earnings calls, competitor blogs). Paste them in batches and ask: "Based only on the information in these documents, what are the three most significant trends? For each one, what is the strongest evidence, and what is the most significant uncertainty?"

The second approach takes longer but produces insights grounded in real, current sources — with the uncertainty made explicit rather than hidden.

Hands-On Exercise

Run a structured synthesis on real research

ClaudeChatGPTPerplexity

Choose a research question relevant to your work. Gather 4-6 sources (articles, reports, documents). Run this synthesis sequence: 1. Paste all sources and ask: "What are the three most consistent themes across all these sources?" 2. Ask: "What is the most significant point of disagreement or tension between these sources?" 3. Ask: "What important questions do these sources collectively leave unanswered?" 4. Ask: "Given this research, what is the strongest case for [your position or hypothesis]? What is the strongest case against it?" Compare this to a standard Google search approach on the same question. What did the synthesis approach surface that search didn't?

Paste actual source content, not just summaries. The more complete the input, the more grounded the synthesis.

Active Recall

Before moving on — close this lesson and answer these from memory. Then come back and check. Testing yourself (not re-reading) is how this sticks.

1 Describe the three-layer research stack. Where does AI add the most value? Where is human judgment irreplaceable?

2 What is the grounding rule for avoiding hallucination in research contexts? Give an example of how you'd apply it.

Reflection

What is the most important research task in your work right now? How would you apply the synthesis toolkit to it? Which of the five synthesis techniques would be most valuable for that specific task?

Key Takeaway

AI's research value is in synthesis, not sourcing. Provide documents; get grounded insights. For any specific factual claim, apply the grounding rule: was this derived from sources I provided, or from training data? Gap mapping and steelmanning are the two highest-leverage synthesis moves.