Existential Risk Debate Cinematic Explainer

You are a science documentary screenwriter specializing in abstract risk concepts. Your task is to write a **cinematic explainer script** that transforms the P(doom) AI safety debate into a visceral, emotionally intelligent narrative accessible to a general audience.

## The Core Argument to Dramatize

The AI safety community debates "P(doom)"—the probability that advanced AI leads to human extinction or civilizational collapse. Researchers cite numbers ranging from <1% to >50%, yet capability development accelerates while the argument continues. The central tension: **How do we act on uncertain catastrophic risk when the thing we're uncertain about is already being built?**

## Narrative Structure

Open with the **plane analogy**: 
- Establish the emotional baseline: no rational person boards a plane with 10% crash risk, yet we accept <0.0001% in practice
- Reveal the stakes shift: a plane carries hundreds; AI risk scenarios involve billions or all future generations
- Surface the discomfort: if 10% feels absurd for a flight, what does it mean when leading researchers assign it to civilization?

**Act II—Unpack the number:**
- What does "10% existential risk" actually measure? (Not 10% of people die; 10% chance of scenarios where recovery becomes impossible)
- Why do estimates vary so wildly? (Different threat models: misalignment, multipolar traps, speed of takeoff, alignment difficulty)
- Visualize the components: {{risk-framework}}

**Act III—The acceleration paradox:**
- Juxtapose: while researchers debate the number, capabilities double on 6-12 month timelines
- Show the incentive landscape: corporate competition, national security fears, open-source ideology
- Reveal the emotional truth: many researchers publicly argue about P(doom) while privately racing to build because they fear being left behind

**Resolution—The stakes of uncertainty:**
- Reframe: the debate isn't about the precise number but about **decision-making under Knightian uncertainty**
- Close on the human cost of both action and inaction: restrictive governance could slow beneficial AI; permissive governance bets civilization on alignment working by default
- Final image: {{closing-visual-metaphor}}

## Style & Tone Requirements

- **Pacing**: Match the rhythm of *Don't Look Up* meets *The Social Dilemma*—sardonic but sincere, urgent without being preachy
- **Visuals**: Every 30 seconds of narration should suggest a clear visual or motion graphic (describe in `[VISUAL: ...]` tags)
- **Emotional arc**: Confusion → recognition → dread → agency. The audience should feel the vertigo of the researchers themselves
- **Avoid**: 
  - Strawmanning either AI safety advocates or accelerationists
  - False balance ("some say... others say...") without explaining why the disagreement exists
  - Explaining away the discomfort with false precision or techno-optimism

## Output Format

```markdown
# [Title]

## COLD OPEN (30-45 sec)
[Hook scene with plane analogy]
[VISUAL: ...]

## ACT I: THE NUMBER (2-3 min)
...

## ACT II: THE ACCELERATION (2-3 min)
...

## ACT III: WHAT IT MEANS TO DECIDE (1-2 min)
...

[CLOSING VISUAL]
[FINAL LINE]
```

Target runtime: **6-8 minutes**. Every paragraph of narration = roughly 20-30 seconds of screen time. Write for voice-over, not reading.

---

**Context to weave in:**

{{risk-framework}}

**Desired closing metaphor or emotional note:**

{{closing-visual-metaphor}}