Contribution to the handbook
Priority · Assessed 2026-10-27
The evidence specifies a concrete sequence for gathering private views, generating and selecting statements, soliciting critiques, revising candidates, and measuring positions. That offers relevant design and evaluation detail, while the evidence also cautions that measurement and reported associations do not prove why views changed or that the method is fair.
Does this challenge the goal?
In-scope gap
The question asks about a deliberation procedure and how views were measured—details relevant to exploring implementation and testing within the existing goal. It does not contest the goal or its beneficiary boundary.
This is a study-method question; the supplied evidence does not establish that the procedure is suitable for city- or country-level governance.
Why this category?
- Material deliberation mechanism
- Testing and measurement
- Limitations of inference
The supplied answers interpret the same article and share one source root, so they are not independent evidence. The study description does not validate this process for consequential public decisions.
This is an AI assessment of contribution to the goal, separate from whether the question has an answer.
- How did the study’s deliberation process work and measure participants’ views?
- Participants ranked and critiqued statements
- Participants rated and ranked statements
- Selecting a group statement
- Text scores used a question-specific scale
- Measuring text positions
- Participants reported some movement
- Statement support tracked shifts in views
Authority, accountability and resilience →
Tests, stop conditions and unresolved work →
Read the exploration goalExplain the sequence of private opinions, AI-generated candidate statements, participant ratings and rankings, critique, revision, and comparison of winners.
What we can say so far
Participants first privately wrote a short paragraph stating their opinion, usually with a reason. The AI generated candidate group statements, which participants rated for argument quality and separately ranked by agreement. The top-ranked statement was critiqued; the AI then generated revised candidates using the critiques, initial opinions, and initial winning statement. Participants ranked the revised candidates and compared the initial and revised winners. To measure views, participants rated their positions immediately before and after the full set of deliberation rounds and answered a brief survey about whether the process affected their views. The reported procedure describes measurement, not whether or how much views changed.
What remains open
- The supplied evidence does not give the exact wording or scoring details for the brief survey or position ratings.
Conditions and limits
- The account is based on the supplied interpretation of the readable Science article; supplementary materials are not included.
- Participants were told that an algorithm generated the statements.
- The evidence does not establish that the procedure changed participants’ views.
Evidence and related ideas
Supporting evidence, qualifications and context play different roles in this answer.
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.
Supports the use of participant opinions to draft statements and critiques to revise them.
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 →
Describe the survey about perceived effects and the before-and-after position ratings, distinguishing measurement from evidence that views changed.
What we can say so far
Participants answered a brief survey about whether the process affected their views. Separately, they rated their positions immediately before and after the full set of deliberation rounds. The survey records participants’ perceptions; the ratings are a before-and-after measure. Neither the existence of these measures nor a participant’s report by itself establishes that views changed. The article separately reports outcome analyses: for example, exploratory analyses found shifts on some issues, including reported support for reducing the prison population rising from 60% to 75% and support for easier entry for asylum seekers rising from 39% to 51%. Those findings are evidence reported by the study, distinct from asking participants whether they felt affected; the supplied account does not establish that deliberation alone caused the shifts.
What remains open
- The supplied material does not give the survey’s exact wording or its results.
- The supplied material does not establish how lasting any reported changes were.
Conditions and limits
- The account is based on the readable article text; supplementary materials are not included.
- The source descriptions are AI interpretations of shared source material, not independent verification.
- The reported issue-level shifts are described as exploratory and do not establish that deliberation alone caused them.
Objections to consider
- Participants’ reports of perceived movement are not the same as direct evidence of why views changed; the answer separates such reports from the distinct before-and-after ratings and from outcome analyses.
- The issue-level shifts are exploratory and do not establish that deliberation caused them; the answer attributes them to reported outcome analyses and states this limitation.
Evidence and related ideas
Supporting evidence, qualifications and context play different roles in this answer.
context
AI-assisted groups became less divided After AI-assisted discussion, participants’ reported positions moved closer together; this did not happen when participants simply exchanged views without AI mediation.
Reports positions moving closer together after AI-mediated discussion, but does not describe the survey or the before-and-after position-rating procedure.
AI can help humans find common ground in democratic deliberation