Smarter-than-human AI as the likely default cause of human extinction, making alignment the most important problem in history.
76
fitness index
Fear-borne and sticky; every capability jump feeds the dread.
FIG. A— genome
plate
Contagion80
Resistance78
Virulence75
Adaptability55
Network Gain82
Emotional Pull90
§ICore Architecture
drives EMO
Central Concept
Smarter-than-human AI is likely to arrive soon and, absent a solved alignment problem, is the default cause of human extinction — making it the most important problem in history
Key Narratives
The misaligned superintelligence that pursues its goals to our ruin, the "p(doom)" probability everyone must now estimate, humanity summoning something it cannot control
Invariant Elements
Intelligence as supremely dangerous, alignment as unsolved and urgent, extinction as the stakes
Obs. in the wild —Yudkowsky-style arguments that a sufficiently capable, misaligned optimizer treats humanity as raw material — illustrated by the "paperclip maximizer"
§IITransmission Mechanisms
drives CON
Primary Vectors
LessWrong and EA forum essays, long podcasts, open letters, mainstream-press profiles, viral "p(doom)" interview clips
Required Resources
A community of careful arguers, a vocabulary (alignment, orthogonality, instrumental convergence), credentialed spokespeople
Ease of Transmission
High once seeded — fear is sticky, and "what's your p(doom)?" is an irresistibly shareable prompt
Obs. in the wild —A researcher casually stating a personal p(doom) on a popular podcast and the number ricocheting across social media for days
§IIISelection & Defense
drives RES
Environmental Factors
Thrives amid each visible AI capability jump, when the gap between progress and understanding feels widest
Defense Mechanisms
Hard to falsify on its own terms — the absence of catastrophe is read as "not yet," and skeptics are cast as dangerously complacent
Competitive Advantage
Confers maximal moral seriousness; if correct, no other cause matters, which crowds out competing priorities
Obs. in the wild —When AI behaves safely, adherents note that current systems simply aren't capable enough yet, so reassuring evidence rarely counts against the thesis
§IVEvolution Pattern
drives ADP
Mutations
From niche LessWrong threads to Bostrom's Superintelligence to mainstream pause letters, governmental safety institutes, and the doomer–accelerationist culture war
Adaptation
Re-anchors its timelines and threat models to each new model release, keeping the threat perpetually near
Stability
The core claim — capability without alignment equals catastrophe — persists while specific scenarios and dates update
Obs. in the wild —Each frontier model launch spawning fresh think-pieces re-dating "when" rather than re-examining "whether"
§VSocial Network Effects
drives NET
Feedback Loops
Shared dread bonds the community, funds research and orgs, which produce more warnings, which deepen the dread
Group Identity
Identity as the clear-sighted few who take the stakes seriously while the world sleepwalks
Status Dynamics
Status accrues to those who saw it earliest and argue the threat most rigorously; a higher, well-defended p(doom) can read as sophistication
Obs. in the wild —EA and rationalist circles treating early, vocal alignment worriers as prescient elders
§VIEmotional Hooks
drives EMO
Primary Emotions
Existential dread, urgency, the grim significance of guarding humanity's survival, in-group solidarity
Reward Mechanisms
The meaning of working on the "most important problem," belonging among the few who understand, relief through action
Motivation
To prevent the end of the world and to matter maximally by doing so
Obs. in the wild —Young researchers redirecting their entire careers to alignment, describing it as the only thing that could possibly matter