Contribution to the handbook
Priority · Assessed 2026-10-28
The question connects citizen authority, government oversight, AI influence on decisions, and capture risks. It can materially inform the division of authority and accountability across governance functions, although its supplied questions and perspectives do not settle those arrangements.
Does this challenge the goal?
In-scope gap
Tests whether people and governments can influence AI development and use, and by what means, within the handbook’s citizen-service and resilience goals. It does not claim that they retain complete control.
The packet raises both formal and practical influence concerns but does not establish how much control people or governments currently exercise or which arrangements are effective.
Why this category?
- Practical and formal ways citizens and governments may shape AI development and use
- Risks that assistance, expert control, or agenda-setting can diminish meaningful public influence
- Unresolved authority boundaries and accountability implications
Some supplied records are questions or perspectives, not evidence of actual control or loss of control. The source passages on influence and risks do not independently establish the effectiveness of any safeguard.
This is an AI assessment of contribution to the goal, separate from whether the question has an answer.
- Do people and governments still have a role in shaping AI?
- Before an AI-mediated account of citizen disagreement informs a binding mandate, how should its fidelity, unresolved objections, and decision-making provenance be checked, and who must authorize any policy choice the account leaves open?
- Citizens authorize rules; AI acts within their mandate
- Which governing decisions may AI make under a citizen-approved mandate?
- How can citizen control coexist with workable oversight?
- When can AI assistance become de facto governing authority without holding formal final-decision power, and what safeguards could preserve democratic control over agenda-setting, available options, implementation, and expert review?
- Better performance, less understanding
- From agents to recipients
- Provider control creates risks
- Reasons for choices are unclear
- Societies can shape AI
- Public input in AI development
Making and interpreting laws →
Judging situations and resolving disputes →
Authority, accountability and resilience →
Tests, stop conditions and unresolved work →
Read the exploration goalAddress whether people and governments can still influence AI development and use, and distinguish their respective means of doing so from any claim that they fully control it.
What we can say so far
Yes, the supplied material presents people and governments as still able to influence AI, but it does not establish that they fully control its development or use. Landemore argues that societies can shape AI and proposes more inclusive democracy as a way to give governments legitimacy and knowledge for oversight. The proposed means include citizen input alongside expertise and regulation, public voice in international institutions, and using AI to strengthen collective democratic governance. A chapter also reports that Polis gathered public input for guidance to Claude 3 and that OpenAI worked with the Collective Intelligence Project on public-input initiatives; those reports show efforts to involve the public, not that public input controlled the resulting systems. The material therefore supports a qualified claim about possible and reported avenues of influence, not a demonstrated account of how much influence they achieve. It does not support a claim of complete control.
What remains open
- The material does not specify the concrete powers, procedures, or institutional channels through which governments can influence particular AI development and use decisions.
- It does not establish how much influence citizens or governments actually have, how their input is weighed, or whether it changes AI systems or their use.
- It does not establish that citizens or governments can fully control AI, nor does it provide a settled boundary between public authority and delegated implementation.
Conditions and limits
- Landemore's claims about political ability and inclusive-democratic oversight are attributed arguments and proposals, not demonstrated results; the source excerpts expressly limit the claimed ability to the present, at least for now.
- The reported public-input initiatives establish that activities were reported, not that public input determined model development, controlled model behavior, or made systems safer.
- The citizen-authority account and citizen-controlled AI-helper approach are proposals; the supplied material does not establish that either is implemented or effective.
- Danaher's concerns about reduced understanding and participation are presented as risks, not established effects in every use of AI.
Objections to consider
- The warning that AI partnerships could reduce people to recipients rather than agents qualifies the claim that people can meaningfully shape AI-related decisions. The supplied material does not establish whether this risk occurs or how to prevent it.
- The chess example challenges the inference that human-AI collaboration or better performance necessarily gives people meaningful understanding or influence. The source itself says the example does not establish a general effect, and no supplied evidence resolves the concern.