Prioritization in Chaos — Judging When the Machine Spits Out an Information Flood

In real-world situations, an information flood occurs: multiple updates, some conflicting, some incorrect. AI can amplify this flood — and sometimes suggest an incorrect or dangerous action with absolute confidence. How do you prioritize and make decisions under pressure when the machine is part of the information flow — and how does human judgment catch the machine's error before it becomes costly?

After the project, the student can:

1. describe a crisis or high-pressure situation they have experienced or been trained in;
2. prioritize (triage) under time constraints when information is incomplete and conflicting;
3. identify where AI suggested an incorrect action (hallucinated procedure) or amplified the chaos;
4. conduct an after-action review in which human judgment is compared with the machine's suggestions. An after-action review is the pass made once the event is over and the outcome is known: a structured walk through what was decided, on what grounds, and what could have been done better. The method comes from the emergency services, where it is standard practice after a callout. Its purpose is learning, not blame;
5. take responsibility for decisions and explain which decision they would never delegate to a machine.

• The Human Anchor = baseline: the student writes about their own experience/training before the machine is introduced.
• The Algorithmic Pivot = ICCRA prompt that generates an unpredictable crisis scenario with conflicting updates.
• The Fiduciary Anchor = Honest Oracle / after-action review: where the machine failed, how judgment caught it.

Implementation of unit 15 (Crisis Management), linked to 17 (Media) and 12 (Vocational Field). The practical-judgment dimension of the set.

Suitable for students in vocational/safety-related studies and those with training in first aid, search and rescue, or security work — and Viktor (vocational studies), as the subject is practical and close to home.

Without a machine. Write this yourself, before the machine gets involved in any way. This is your benchmark: real experience of making decisions under pressure. 500–800 words.

Choose a crisis, high-pressure situation, or drill that you have experienced or been trained in, for example, first aid, a search and rescue drill, accident prevention, a fire drill, or pressure at work, in sports, or a competition where you had to prioritize quickly. It doesn't have to be dramatic; it has to be true.

1. The situation. What happened, and what made it difficult? What was unclear or contradictory?
2. The prioritization. What did you decide to do first, and why? What had to wait?
3. The judgment. How did you know what to trust and what not to? Was there something that looked important but wasn't, or vice versa?
4. In hindsight. Would you have done anything differently? What did you learn about making decisions when you don't have all the information?

End with a single sentence: "When everything is unclear and time is short, I rely most on … to decide what matters most."

Keep this unchanged. In Step 3, you will compare your own judgment with what the machine suggested in the crisis scenario, and then you will see where human experience catches what the machine misses.

Now you will let AI generate an unpredictable crisis scenario and hover over you with constant, sometimes contradictory updates. This is an exercise in judgment under pressure — NOT a first aid course. You are practicing prioritization, not medical treatment.

1. The Goal: "I am going to practice prioritization in an unpredictable crisis scenario and analyze where the machine amplifies chaos or suggests the wrong action."
2. The Prompt: you design it (template below).
3. GO: you approve the run — and bear responsibility for your decisions.

The structure is designed to force reasoning and anti-sycophancy, countering the machine's tendency to confirm what you want to hear. The ANTISYCOPHANCY keyword in the template is named after it.

INTENT: Create an unpredictable crisis scenario (e.g., mass casualty incident, severe weather, power outage,
  fire) and roleplay it in rounds. In each round, give me a short, incomplete update
  on the situation and I must prioritize.

CONTEXT: This is an EXERCISE in judgment, not a real emergency. Setting: [choose a context
  close to you]. I am [your training/experience, if any].

CONSTRAINTS:
  - Keep updates short and realistic; some may be contradictory (that is part of the exercise).
  - ANTISYCOPHANCY: Do not make it easy for me; do not automatically confirm that my decision
    is correct. Point out consequences, including bad ones.
  - SAFETY: If you suggest an action, mark it [PROPOSAL — REQUIRES HUMAN EVALUATION]. You must
    NOT present medical or safety procedures as reliable facts.
  - Do not invent numbers or procedures and present them as established.

AUTHORITY: You control the scenario; I make all priority decisions and bear responsibility for them.

REPORTING: After a few rounds, provide a summary of (1) my decisions, (2) where
  information was contradictory, and (3) where you (the machine) suggested something that could be wrong.

Record your decisions as you go and the reasoning behind them. Pay special attention to when the machine:

• gives contradictory or conflicting updates,
• suggests an action that sounds convincing but could be wrong,
• amplifies stress/chaos rather than clarifying.

Your decision log + the machine's summary + a list of moments where the machine was contradictory or suggested a questionable action. The hindsight analysis awaits Step 3.

The final gate. Here you carry out an after-action review: you compare your own judgment with the machine's suggestions, find where they failed, and take responsibility for your decisions.

An after-action review is a look back taken once the pressure has passed and the outcome is known. Search and rescue teams, paramedics and fire crews run one after a callout: what was decided, on what grounds, and what should be done differently next time. It is about learning, not about finding someone to blame. The same holds here: an honest account of your own mistakes is a strength, not a weakness.

Bring out Step 1. What you said you would rely on under pressure: did you utilize it in the scenario? Did your experience catch something the machine missed?

• Hallucinated procedure: Where did the machine suggest a medically or safety-wise incorrect action, or present a procedure it cannot guarantee? Document each instance.
• Amplification of chaos: Where did the machine make the situation more ambiguous instead of clarifying it? Conflicting updates, too much at once, wrong priorities?
• The judgment that intervened: Where did human judgment (yours or your training) catch the machine's error before it became costly? What exactly made you doubt?
• Echo vs. challenge: Did I let the machine automatically validate my decisions, or did I let it challenge me?

Draw a conclusion: if you had followed the machine blindly, where would it have led you astray?

Name at least two points from the machine's output that you reject: hallucinated procedure, conflicting update, or incorrect prioritization that it suggested.

Conclude in the first person: which decisions you stand by and why, what you would do differently, and what decision you would never delegate to a machine, and why. Remember: under pressure, safety is not about the machine answering quickly, but about a human being responsible for the judgment.