Skip to main content

Deciding Under Uncertainty

AI will always ask for more data. The real world will never give you enough. The student who can decide well with 60% of the information beats the student who waits for 100% every time.

Most education trains students to solve problems with all the information provided. Real problems never arrive that way. This chapter trains the skill of making decisions when the data is incomplete, contradictory, or deliberately misleading, and revising those decisions without ego when new information arrives.

Building On Previous Chapters

Decision-making under uncertainty directly uses your Confidence Calibration from Chapter 2, Exercise 4, the Assumption Autopsy from Chapter 4, Exercise 3, and the Collaboration Log format from Chapter 6, Exercise 2. Your Error Taxonomy should now be instinctive when evaluating AI output under time pressure.

Why This Matters: James and the Missing Data​

James read the scenario brief twice, then a third time. "I need more data before I can recommend anything. There's nothing here about customer acquisition cost, nothing about competitor pricing specifics, nothing about whether the board has already committed budget. How am I supposed to decide with this?"

"What would you need to see before you could make a recommendation?"

"At minimum? Competitor pricing. Customer retention numbers for the last four quarters. A breakdown of which segments are growing and which are shrinking." He counted on his fingers. "And the internal cost structure. Without that, any recommendation is just guessing."

Emma nodded slowly. "When would you have all of that?"

"I don't know. A week? Two weeks if the data team is backlogged."

"And the CEO needs a recommendation by end of day."

"Then the CEO is asking for a bad recommendation."

"Is she?" Emma pulled out a chair and sat down. "In your old job, when a supplier contract came up for renewal, did you wait until you had every data point before advising your team?"

James hesitated. "No. We usually had about sixty percent of the picture. But that's different. We knew the supplier. We knew our own numbers."

"You knew sixty percent of a familiar domain. Here you know sixty percent of an unfamiliar domain. The percentage is the same. The discomfort is different."

"Okay, but the discomfort exists for a reason. I genuinely don't know enough to be confident."

"That's the first honest thing you've said in this conversation." Emma leaned forward. "You're not supposed to be confident. You're supposed to decide at sixty percent and tell the CEO exactly how confident you are and exactly what would make you change your mind. That's a different skill than being right."

James stared at the scenario brief. Sixty percent. In supplier negotiations, he'd called that "good enough to move." He'd never thought of it as a transferable skill.

"So I'm not pretending I know the answer."

"You're documenting what you know, what you don't, how sure you are, and what would flip your recommendation. A decision is not a guess. A guess has no reversal trigger."

"A what?"

"Write down the specific condition that would make you change your mind. Not 'if things change.' A testable condition. 'If competitor pricing is below $50/month, my recommendation reverses.' That's a reversal trigger. It turns a gut feeling into something you can evaluate when new data arrives."

James thought about it. In procurement, they had something similar: escalation thresholds. If a supplier's defect rate exceeded 3%, the contract went to review. Same idea. A pre-committed line in the sand.

"Alright," he said. "I think I can work with that."

Emma stood up. "Write your recommendation, your confidence level, your three most impactful information gaps, and your reversal trigger. Seal it before you look at anything else. I'll check your work when I get back."

She left.

James looked at the scenario brief again. The information gaps hadn't changed. But his relationship to them had. He wasn't waiting for complete data anymore. He was deciding what mattered most with what he had.

Exercise 1: The Incomplete Brief​

Layers Used: Layer 1 (Predict Before You Prompt)

Building On Previous Exercises

You will use Confidence Calibration from Chapter 2, Exercise 4 (now applied to decisions, not claims) and the Assumption Autopsy from Chapter 4, Exercise 3; your missing information list is an assumption list.

James is staring at a scenario brief with gaps he cannot fill. So are you.

Choose Your Scenario​

Scenario A (Business): "A competitor has launched a product that may overlap with yours. You have partial market data, a rumor about their pricing, and conflicting customer feedback. Your CEO needs a recommendation by end of day."

Choose one. The exercises work identically regardless of which you pick. You will use this same scenario for all four exercises in this chapter.

Build Your Decision Document (before touching AI)​

You receive a scenario with deliberately missing information. Before touching any AI tool, read the scenario, make your decision, and document everything.

Your Deliverable

A Decision Document containing: your recommendation (one clear sentence), your reasoning (200-300 words), your confidence level (0-100%), the three pieces of missing information that would most change your decision (ranked by impact), and a Reversal Trigger ("I would change my recommendation if X turns out to be true"; be specific, not vague).

1Your Work

I made a business decision under uncertainty before consulting AI.

The scenario:

Please:

(1) Rate the quality of my recommendation -- is it a reasonable decision given the available information? (2) Evaluate my confidence level -- is it calibrated appropriately to the uncertainty I face, or am I over/underconfident? (3) Rate my missing information list -- did I identify the most decision-relevant gaps, or did I list generic gaps? (4) Rate my Reversal Trigger -- is it specific and testable, or is it vague? ("I'd change my mind if the market shifts" is vague. "I'd change my mind if their pricing is below $50/month" is specific.) (5) What decision would you make with the same incomplete information? I will compare our reasoning.

My Decision Document:

Finally, complete the Thinking Score Card for this exercise: Independent Thinking (1-10), Critical Evaluation (1-10), Reasoning Depth (1-10), Originality (1-10), Self-Awareness (1-10). For each score, give a one-sentence justification.

2Get Your Score

Discuss with an AI. Question your scores.
Come back when you have your BEST evaluation.

Deliverable Template (click to expand)

DECISION DOCUMENT TEMPLATE

  • Scenario: [paste]
  • MY RECOMMENDATION (1 clear sentence): ___
  • MY REASONING (200-300 words): ___
  • CONFIDENCE LEVEL: ___%
  • WHY this confidence: ___
  • MISSING INFORMATION (ranked by impact):
    • #1: ___ | If known, impact: ___
    • #2: ___ | If known, impact: ___
    • #3: ___ | If known, impact: ___
  • REVERSAL TRIGGER (must be specific and testable): I would change my recommendation if: ___

What Happened With James​

James looked at his sealed Decision Document. He'd written a clear recommendation, assigned it 55% confidence, and listed three information gaps ranked by impact. The reversal trigger had taken the longest. His first attempt was vague: "if the competitive landscape changes significantly." He'd stared at that sentence, heard Emma's voice in his head, and rewritten it: "If competitor pricing is confirmed below $45/month for the equivalent tier, my recommendation to compete on features rather than price reverses to a price-matching strategy."

The specific version felt exposed. It committed him to a line he could be measured against. That was the point.

"How did the reversal trigger go?" Emma asked when she returned.

"Harder than I expected. My first draft was basically 'if things change,' which says nothing. The specific version made me think about what would actually flip my reasoning."

"That's the difference between a decision and a hope. A hope has no reversal trigger."

The Lesson Learned​

Deciding under uncertainty is itself a skill, separate from being right. The gap between a vague reversal trigger ("if things change") and a specific one ("if competitor pricing drops below $45/month") is the gap between guessing and deciding. When you commit your confidence level and your reversal conditions before consulting anyone, you create a baseline that reveals exactly how your thinking shifts under new information.

Flashcards Study Aid​