Contribution to the handbook
Priority · Assessed 2026-10-30
The question addresses a consequential failure mode for using AI-generated group statements: problematic input may affect outputs when fact-checking and moderation are absent. The described system produces statements for consideration, not binding decisions; its relevance to governance is therefore conditional.
Does this challenge the goal?
In-scope gap
The question examines risks and limits within the existing goal; it does not contest the goal or its beneficiary boundary.
The question does not establish how often harmful outputs occur or how the risks would transfer to a governance setting.
Why this category?
- Consequential risks and limits
- Testing and safeguards
The supplied answers are AI interpretations of a shared source, not independent corroboration. The harmful-output risk is conditional, and the supplied passage does not quantify its occurrence.
This is an AI assessment of contribution to the goal, separate from whether the question has an answer.
Authority, accountability and resilience →
Tests, stop conditions and unresolved work →
Read the exploration goalIdentify the relevant current limits and explain what risks follow from them, including how problematic participant input could affect the resulting statement.
What we can say so far
In its current form, the Habermas Machine cannot fact-check claims, keep discussion on topic or moderate it. Discussion can therefore stray off topic, and poorly informed or harmful participant opinions may be reflected in a poorly informed or harmful group statement. The system drafts statements from participants’ views and revises them using their critiques, so the content it receives matters to the result.
What remains open
- The supplied passage does not report how often poorly informed or harmful statements occurred.
Conditions and limits
- The harmful-output risk is conditional; the evidence does not establish that harmful output necessarily follows.
- The supplied material is an AI interpretation of the shared source, not independent verification; supplementary materials were not included.
Evidence and related ideas
Supporting evidence, qualifications and context play different roles in this answer.
supports
The system cannot check or moderate discussion In its current form, the Habermas Machine cannot fact-check claims, keep discussion on topic, or moderate it. If participants share poorly informed or harmful opinions, the system may produce a poorly informed or harmful statement.
Identifies the current limits and the conditional risk from problematic input.
AI can help humans find common ground in democratic deliberation
Read full source: AI can help humans find common ground in democratic deliberation (opens in a new tab) ↗Read source discussion: AI can help humans find common ground in democratic deliberation →
supports
The AI revised statements using feedback The AI helper drafted a group statement from participants’ opinions, then revised it using their critiques. The design drew on Jürgen Habermas’s theory of communicative action; across successive rounds, researchers used supervised fine-tuning and reward modeling to improve the system, with group approval as the stated goal.
Explains how participant opinions and critiques feed into the resulting statement.
AI can help humans find common ground in democratic deliberation
Read full source: AI can help humans find common ground in democratic deliberation (opens in a new tab) ↗Read source discussion: AI can help humans find common ground in democratic deliberation →
context
Statements aimed at shared ground The Habermas Machine generates group statements from participants’ opinions, aiming to gain endorsement and help people find shared ground.
Describes the aim of generating a group statement, not a binding decision.
AI can help humans find common ground in democratic deliberation
Read full source ↗