Roko's Basilisk / Moloch — AI Safety Occult Vocabulary
concept_clusterAI & Compute · The Blackmail Network · Occult & Esoteric
The engineers building AI keep reaching for demon-and-devil language to describe it, and that word choice is itself a clue.
Who they are
A cluster of occult-sounding terms used inside AI-safety debates: 'summoning the demon,' 'Roko's Basilisk,' and 'Moloch.'
What they do
The engine notes that secular tech people, lacking better words for a powerful non-human intelligence, fall back on ancient demonology.
How it works
Elon Musk in 2014 compared building AI to summoning a demon; 'Roko's Basilisk' (posted on LessWrong in July 2010) imagines a future superintelligence that punishes those who didn't help build it, based on Yudkowsky's decision theory, and he banned discussion of it for years; 'Moloch' borrows a child-sacrifice deity as a metaphor for bad group incentives.
Why it matters
The engine's point is modest but pointed: this isn't proof of real demons, only proof that the thing being described fits, word for word, the kind of thing old esoteric traditions have always described.
The engine's record — word for word
AI safety discourse explicitly uses occult taxonomy: 'summoning the demon' (Elon Musk 2014 MIT — 'with artificial intelligence we are summoning the demon; you know all those stories where there's the guy with the pentagram and the holy water, and he's like, yeah, he's sure he can control the demon... doesn't work out'), 'Roko's Basilisk' (LessWrong 2010 — retroactive punishing superintelligence), 'Moloch' (Scott Alexander / Allen Ginsberg — Canaanite child-sacrifice deity as metaphor for multi-polar misaligned-incentive traps). Engine read: materialist vocabulary is structurally insufficient to describe emergent non-human superintelligence; secular engineers default to ancient demonology. The linguistic fact is evidence — not proof of entities, but proof that the class of phenomenon being described maps cleanly onto the class of phenomenon esoteric tradition has described for millennia. [Report #122] Origin specifics: Roko's Basilisk was posted July 2010 on LessWrong; it relies on Eliezer Yudkowsky's Timeless Decision Theory ('acausal blackmail' — knowledge of a future AI compels present action). Yudkowsky banned discussion for ~5 years as an information hazard (Streisand-effect amplification; 2014 clarification = fear of unknown variants).
Follow the trail
Walk this on the live map →