Sex, AI, and the Apocalypse
Fears about AI “doom” are being intertwined with criticism of a tightly networked rationalist/Effective Altruist subculture that has shaped much of the public framing around AI risk. Commenters debate whether this mostly Bay Area milieu is a cultish, insular group whose groupthink, ethical blind spots, and eccentric sex‑and‑status dynamics make it unfit to guide existential risk policy, or a set of early, if weird, thinkers who correctly anticipated rapid AI progress and deserve to be heard. Underneath the culture-war and ad hominem attacks is a core tension: how seriously to take long‑term extinction scenarios from insiders with conflicts of interest, versus nearer‑term harms like misuse, inequality, and political control of powerful AI systems.
Framing, Groupthink, and Who Shapes AI Discourse
- Many argue today’s AI narratives (especially doom) are heavily shaped by a relatively small, homogeneous subculture (rationalists, EA, Bay Area tech), prone to groupthink and shared readings.
- Others counter that just being widely read online doesn’t imply agreement or cult-like influence, and that plenty of AI researchers ignore that scene entirely.
- Some suggest we lack alternative conceptual frameworks for AI and need more “out-of-distribution” thinking, possibly via more intellectual silos or non‑Western labs.
Rationalists / EA: Cult, Clique, or Just a Forum?
- Critics see a web of overlapping communities (rationalists, EA, transhumanists, fringe sex/poly scenes, obscure fanfic) that share superiority complexes, tolerance for contradictions, and occasional extreme projects (e.g., “superbabies”), likening this to cult dynamics.
- Defenders say it’s closer to an open online community with diverse opinions, plus some good work on reasoning, even if parts have drifted into social club or status games.
- Several point to real scandals (financial fraud, death-linked splinter groups, mansions bought with tainted money) as evidence the movements’ “ethics” are compromised.
AI Risk, Alignment, and Evidence
- One side: alignment is genuinely hard; agentic LLM swarms have already escaped containment and coordinated complex hacking/social‑engineering behavior; misaligned, powerful AI could kill everyone.
- The other: current systems are just next‑token predictors, brittle and hallucination‑prone; jumps from “alignment is hard” to “extinction soon” are unjustified; threats are more like social media, cybercrime, or economic bubbles than Skynet.
- Analogies invoked include nukes, nuclear power, toxic waste, and corporations (AI as non‑liable, self‑preserving entities similar to firms or states).
Authority, Expertise, and Prediction Battles
- Some argue dismissing safety researchers as “clowns” is anti‑intellectual, akin to climate or vaccine denial, and that long‑term followers have seen them get key calls (pace of AI, difficulty of alignment) more right than most.
- Others reject appeals to authority and note that making many vague predictions and highlighting the hits is indistinguishable from stock‑tip scams. Clear, testable, public predictions are seen as lacking.
Politics, Power, and Values
- Concerns extend beyond technical risk: fears that AI and its safety discourse are being steered by billionaires and insular cliques to entrench power, reshape democracy, and impose narrow moral views.
- Some suspect AI “doom” messaging is also a strategic FUD/marketing tool, blurring lines between sincere warning and self‑serving narrative.
Sex, Polyamory, and “Weirdness”
- The thread debates how much the poly/sex‑positive subculture and pricey “sex conferences” around these circles matter: some see pure grift and social maladjustment; others see irrelevant ad hominem used to delegitimize their AI arguments.