Could people accept language-based machine processes as a basis for shared decisions?

Contribution to the handbook

Priority · Assessed 2026-10-30

Public acceptance is a consequential condition for using language-based processes in shared decisions. The chapter raises a specific concern about understanding and accepting machine dictation; the study of AI-mediated statements offers relevant but limited evidence, not acceptance of a governing process.

Does this challenge the goal?

In-scope gap

The question asks about acceptance of a particular decision process, rather than contesting the exploration’s goal. Public acceptance and its conditions remain open within the existing definition.

Acceptance could affect whether a proposed arrangement is desirable, but the question does not establish that people accept or reject AI governance.

Making and interpreting laws →

Authority, accountability and resilience →

Tests, stop conditions and unresolved work →

Assess the possibility of public acceptance, including relevant reasons for confidence or concern, without presenting a proposed process as already accepted.

What we can say so far

Acceptance is possible, but broad public acceptance of a language-based process for shared decisions is not established. A study cited in the augmented-democracy preprint reports that over 60% of its participants would accept a process assigning AI 25% influence and politicians 75%; this was stated willingness, and the excerpt does not establish how broadly it applies or whether the AI was used in the language-based process at issue. In a separate AI-mediation study, group agreement increased and participants preferred AI-written statements on endorsement ratings even after controlling for length. Agreement with a statement, however, is not the same as accepting a decision-making process. There are also reasons for concern. The Plurality chapter suggests that people may struggle to understand or agree on how language input becomes machine dictation. Danaher argues that opaque systems can threaten participation, understanding and legitimacy; the Habermas Machine authors note that some people object in principle to computers generating political or social ideas. These are concerns and normative arguments, not measurements of how many people would reject a particular process.

What remains open

• Evidence of acceptance of a specified language-based process for actual shared public decisions, including its authority and context.
• Evidence on how acceptance varies with AI authority, decision context, participation, transparency and other safeguards, or across affected groups.
• The prevalence and consequences of the identified objections and which safeguards would be sufficient to address them.