Worldview Profiles

People’s views on AI safety tend to cluster into recognizable worldviews. Each worldview combines beliefs about risk, feasibility, and strategy into a coherent package.

Major Worldviews

Worldview	Core Belief	P(doom)	Priority Approaches
AI Doomer	Alignment is very hard; timelines short	30-90%	Pause, agent foundations, governance
Optimistic Alignment	Alignment is tractable engineering	<5%	RLHF, evals, lab culture
Governance-Focused	Technical solutions need policy support	10-30%	Policy, compute governance, coordination
Long-Timelines Technical	We have time for careful research	5-20%	Interpretability, agent foundations

How Worldviews Form

Your worldview emerges from your answers to key cruxes:

Short timelines? ──┬── Yes ──► Alignment hard? ──┬── Yes ──► DOOMER
                   │                              └── No ───► Urgent but optimistic
                   │
                   └── No ───► Alignment hard? ──┬── Yes ──► LONG-TIMELINES TECHNICAL
                                                 └── No ───► OPTIMISTIC ALIGNMENT

Governance-focused worldview cuts across these based on belief that technical solutions alone are insufficient.

Using This Section

1. Identify your worldview: Which profile is closest to your current beliefs?
2. Understand alternatives: What would you need to believe to hold a different view?
3. Find cruxes: Where do you differ from other worldviews?
4. Steel-man: Can you articulate the strongest version of views you disagree with?

Cross-Cutting Dimensions

Some beliefs vary within each worldview:

Dimension	Spectrum
AI assistance	Trust AI to help with safety ↔ Keep humans in charge
Open source	Net positive ↔ Net negative
Lab work	Change from inside ↔ Change from outside
Theory vs. empirical	Formal methods ↔ Experimental safety