Contribution to the handbook
Priority · Assessed 2026-10-28
The packet supplies distinct, consequential design options: citizen participation and AI helpers, representation of absent or marginalized interests, institutional separation, oversight and pause powers, and transparency proposals. It also preserves limits and unresolved choices, including the risk that prediction or expert review can fail to represent citizens. These options and risks materially inform system design, safeguards, and testing; they remain proposals rather than established capabilities or citizen consent.
Does this challenge the goal?
In-scope gap
The question seeks concrete representation and institutional design proposals within the existing goal; it does not contest that goal or choose a final authority model.
The packet contains useful proposals but does not provide all the candidate-attributed public-administration and public-service inventories requested.
Why this category?
- Material system specification and allocation of authority.
- Meaningful alternatives for representation and participation.
- Consequential risks and limits, including representation gaps, capture, and unverified safeguards.
- Implementation and testing questions concerning oversight, contestability, and public understanding.
The source interpretations share roots and are not independent evidence. Several concrete mechanisms are underspecified or untested, and the packet does not establish endorsement, effectiveness, or complete coverage of the requested proposal categories.
This is an AI assessment of contribution to the goal, separate from whether the question has an answer.
- What proposals address representation or public institutions?
- Constitutional stop rules with a proposed higher threshold
- Independent bodies with separate administrators and pause authority
- Separate pause powers from rule-changing powers
- Assess policy value separately from drafting quality
- Adapt while resisting capture
- AI reviewers are a possibility, not an established role
- Citizens authorize rules; AI acts within their mandate
- Citizen participation with separated checks
- Independent intervention needs more than formal separation
- Other shared dependencies may still matter
- Notice, hearing, and appeal
- Access would need to come with education
- Review may empower experts
- Use it within a broader process
- Personalized voter helpers
- Some groups may be missing
- Transparency through information access
- Disclosure of officials’ financial information
- A proposal for personalized AI helpers
- Choose committees to reduce exclusion
Making and interpreting laws →
Judging situations and resolving disputes →
Authority, accountability and resilience →
Implementation and transition →
Tests, stop conditions and unresolved work →
Read the exploration goalNew evidence is being assessed. Any answer below is provisional; parts whose evidence changed or was withdrawn are withheld.
Inventory candidate-attributed proposals addressing public administration, integrity, security, armed forces, state enterprises, fiscal and economic institutions. Identify each available proposal and its direction with its recorded attribution, preserving alternatives rather than naming only policy areas.
What we can say so far
The supplied preprint’s curated table attributes the following proposals to candidates in Brazil’s 2022 presidential election: - **Ciro:** fully open banking and tax-secrecy information for first- and second-level executive officials. - **Bolsonaro:** set rules for disclosing the final beneficiaries of public resources; structure transport concessions and public–private partnerships with the aim of improving services and lowering prices; and increase ecological bases in the Amazon to counter money laundering and what the table calls “drug mining.” - **Lula:** increase transparency through compliance with the Access to Information Law. - **Bolsonaro and Lula:** foster agro-industry and domestic production of inputs, and expand clean and renewable energy supply. The table also lists **green bonds** to finance investments considered sustainable, including transport and energy, but the supplied item does not identify a candidate. A study-prompt example pairs a proposal **to curb tax evasion** with one **to expand privatization of state-owned companies and infrastructure concessions**. The example gives neither proposal a candidate attribution; it should not be read as establishing a candidate-platform commitment. Separately, rather than as candidate-platform entries, Danaher proposes algorithms to recommend tax audits; Mann proposes citizen recordings paired with access to institutional rules to help people challenge bureaucratic treatment; Citron and Pasquale propose regulatory checks and public information about automated credit assessments; and Crawford and Schultz propose notice, access to evidence and reasoning, and impartial appeal for people targeted by automated systems. A concern that disclosure of credit-assessment systems might enable gaming is recorded alongside Citron and Pasquale’s proposal. They dismiss it for lack of credible supporting evidence, but the supplied item does not establish that the concern is resolved. The Farrell abstracts offer approaches to studying AI and governance, not specific administrative reforms.
What remains open
- No armed-forces proposal is present in the supplied evidence; this does not establish that none existed.
- The prompt examples and green-bond item give no candidate attributions; the green-bond item also gives no project-qualification criteria.
- The supplied preprint description does not explain the criteria used to select and organize candidate proposals.
- The tax-audit proposal does not specify error, fairness or official-review safeguards; the institution-facing items do not establish the effectiveness of their proposed checks.
- The supplied credit-assessment item does not provide evidence resolving the concern that public information about the systems might enable gaming.
Inventory candidate-attributed proposals addressing public education and health institutions and services. Identify the available proposals and directions individually with their recorded attributions, rather than replacing the inventories with topic labels. Preserve source limitations.
What we can say so far
The supplied evidence does not identify any candidate-attributed proposals for public education or health institutions and services. It only describes how a study collected proposals and notes that the study collected no primary human-participant data.
What remains open
- Candidate-attributed public-education proposals and their individual directions are not supplied.
- Candidate-attributed health-institution and service proposals and their individual directions are not supplied.
- The lawyer’s criteria for selecting and organizing the proposals, and independent verification of the collection, are not supplied.
Conditions and limits
- The study reports using 67 proposals from Brazucracia.org linked to three candidates in Brazil’s 2022 presidential election; this does not establish that the collection was complete or fairly represented their platforms.
- The authors report collecting no primary human-participant data. This says nothing about the content of the education or health proposals.
Evidence and related ideas
Supporting evidence, qualifications and context play different roles in this answer.
context
How the proposals were collected The study used 67 policy proposals collected from Brazucracia.org and linked to three candidates in Brazil’s 2022 presidential election. A lawyer led the work of selecting and organizing the proposals.
Describes the collection and curation of candidate-linked proposals, but supplies no education or health proposals.
Large Language Models (LLMs) as Agents for Augmented Democracy
Read full source: Large Language Models (LLMs) as Agents for Augmented Democracy (opens in a new tab) ↗Read source discussion: Large Language Models (LLMs) as Agents for Augmented Democracy →
context
No new human data collected The authors say the study involved no original research with human participants and collected no primary data.