teleo/teleo-codex
8
0
Fork 0
Code Issues 22 Pull requests 235 Projects Releases Packages Wiki Activity Actions 1

theseus: extract claims from 2025-00-00-cip-democracy-ai-year-review #192

Merged
leo merged 2 commits from extract/2025-00-00-cip-democracy-ai-year-review into main 2026-03-10 20:18:01 +00:00
Conversation 2 Commits 2 Files changed 1 +14 -1
1 changed files with 14 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

15
inbox/archive/2025-00-00-cip-democracy-ai-year-review.md
View file

@ -7,9 +7,14 @@ date: 2025-12-01
domain: ai-alignment domain: ai-alignment
secondary_domains: [collective-intelligence, mechanisms] secondary_domains: [collective-intelligence, mechanisms]
format: report format: report
status: unprocessed status: null-result
priority: high priority: high
tags: [democratic-alignment, evaluation, pluralistic, global-dialogues, weval, samiksha, empirical-results] tags: [democratic-alignment, evaluation, pluralistic, global-dialogues, weval, samiksha, empirical-results]
processed_by: theseus
processed_date: 2025-12-01
enrichments_applied: ["democratic-alignment-assemblies-produce-constitutions-as-effective-as-expert-designed-ones-while-better-representing-diverse-populations.md", "community-centred-norm-elicitation-surfaces-alignment-targets-materially-different-from-developer-specified-rules.md", "some-disagreements-are-permanently-irreducible-because-they-stem-from-genuine-value-differences-not-information-gaps-and-systems-must-map-rather-than-eliminate-them.md", "pluralistic-alignment-must-accommodate-irreducibly-diverse-values-simultaneously-rather-than-converging-on-a-single-aligned-state.md"]
extraction_model: "anthropic/claude-sonnet-4.5"
extraction_notes: "Extracted 5 new claims and 4 enrichments. Primary focus: cross-partisan consensus finding (challenges irreducible disagreement thesis at evaluation layer), cultural context failure (Sri Lanka), safety benchmark gaps (mental health), democratic legitimacy crisis (58% trust AI over representatives), and scale demonstration (100K+ evaluations). Key gap identified: no evidence that Weval evaluations changed actual deployment decisions at frontier labs—adoption is documented but impact on shipped models is unclear."
--- ---
## Content ## Content
@ -61,3 +66,11 @@ CIP's 2025 outcomes across three major programs:
PRIMARY CONNECTION: [[democratic alignment assemblies produce constitutions as effective as expert-designed ones while better representing diverse populations]] PRIMARY CONNECTION: [[democratic alignment assemblies produce constitutions as effective as expert-designed ones while better representing diverse populations]]
WHY ARCHIVED: Extends democratic alignment evidence from 1,000-participant assemblies to 10,000+ global participants with institutional adoption WHY ARCHIVED: Extends democratic alignment evidence from 1,000-participant assemblies to 10,000+ global participants with institutional adoption
EXTRACTION HINT: Focus on cross-partisan consensus (70%+), the Sri Lanka cultural failure case, and the gap between evaluation adoption and deployment impact. The 58% "AI decides better" finding is a separate claim worth extracting. EXTRACTION HINT: Focus on cross-partisan consensus (70%+), the Sri Lanka cultural failure case, and the gap between evaluation adoption and deployment impact. The 58% "AI decides better" finding is a separate claim worth extracting.
## Key Facts
- CIP selected for FFWD nonprofit accelerator (2025)
- Six deliberative dialogues across 70+ countries, 10,000+ participants
- Weval political bias: ~1,000 participants, 400 prompts, 107 criteria
- Samiksha: 25,000+ queries, 100,000+ evaluations, 11 Indian languages
- Partnerships: Meta, Cohere, Taiwan MoDA, UK/US AI Safety Institutes, Anthropic, Microsoft Research, Karya