postgrok

doomer-taxonomy

Properties1
tagsai, taxonomy, policy, discourse

2 min read

Doomer taxonomy (AI harm debate)

Definition: three-camp map of the AI harm discourse — Doomers (pessimists), Accelerators, and Centrists (moderates) — with their intellectual genealogies, motivations, and policy prescriptions.

Origin

2026-10-04-youtube-narayanan-ai-harm-taxonomy — Arvind Narayanan’s video mapping the debate. The confusion in AI policy stems from multiple groups calling for “regulation” with fundamentally different diagnoses.

Three camps

1. Doomers (pessimists)

Lineage: Extropians (1990s SV futurists) → Rationalists (Yudkowsky/LessWrong) → Effective Altruism (MacAskill/Ord/Bostrom, longtermism) → Anthropic, OpenAI Core belief: Superintelligence inevitable; either benevolent god or extinction. Alignment is the central problem. Longtermism: trillions of future lives outweigh current harms. Internal tension: Lab insiders (Anthropic, OpenAI) think they can steer to good superintelligence; external rationalists (Yudkowsky) say labs move too fast and will produce bad superintelligence. Policy influence: Heavy EA funding, congressional fellows placing staffers, shaping Democratic bills toward superintelligence-focused regulation.

2. Accelerators

Composition: VCs (Andreessen, All-In), non-Doomer AI labs (Meta, Google), White House / executive branch Core belief: Doomers are wrong/weird. AI benefits enormous; slowing down is harmful. Race with China requires speed. Doomer ideology traces to transhumanism, not CS. Motivation: Money (VC returns, chip sales) and power (right-coded SV alliances with White House).

3. Centrists (moderates)

Position: Between the poles. Agree with Doomers that guardrails needed; agree with Accelerators that Doomer ideology (superintelligence obsession) is misguided. Three subgroups:

  • AI Ethics: Pre-LLM field. Focus on present/near-term harms: bias, surveillance, discrimination. Early critics of TESCREAL ideologies.
  • Traditional CS / NTCS: Narayanan, Marcus. Evaluate actual tech, skeptical of superintelligence claims, concerned about dangerous agent behaviors from lab race dynamics. Call for ordinary tech regulation.
  • Cybersecurity experts: Focus on LLM hacking capabilities, code generation risks. Practical, grounded threat model.

Shared view: Regulate AI like any powerful technology — specific methods, specific guarantees, not ideological superintelligence frameworks.

Key structural insight

Doomers and Centrists both call for “regulation” but mean opposite things. Congressional staffers primarily hear the Doomer frame via EA lobbying/fellows.

Graph View

Backlinks