Contribution to the handbook
Priority · Assessed 2026-10-28
The reported comparison bears directly on whether model predictions might improve a preference estimate relative to a sample alone. The positive result is bounded to the study data, and matching that sample does not establish representation of people outside it; that limitation materially affects any proposed use.
Does this challenge the goal?
In-scope gap
It asks whether model augmentation improved a group-preference estimate, a test relevant to citizen-input methods within the existing goal, without contesting that goal.
The comparison concerns reproducing the preferences of the study’s participants, not establishing what a wider public prefers.
Why this category?
- alternatives
- tests
- risks
The packet gives a positive reported result for some tested approaches, but not a full account of its size, consistency, or applicability to other populations and decisions.
This is an AI assessment of contribution to the goal, separate from whether the question has an answer.
- Did model predictions improve the group-preference estimate in the reported comparison?
- Testing whether a model improves a sample
- Testing group-level preferences
- The data used in the comparison
- How the study measured a match
- Predictions improved some sample matches
- Small samples improved in this test
- Summarizing group preferences with win rates
- Estimates of group preferences
- Comparing sample and full-sample patterns
Making and interpreting laws →
Authority, accountability and resilience →
Tests, stop conditions and unresolved work →
Read the exploration goalState the reported direction of the comparison between the sample augmented with model predictions and the sample alone for estimating group preferences.
What we can say so far
Yes. The reported comparison found that samples augmented with model predictions estimated the study group’s aggregate preferences more accurately than samples alone. The result is reported for the study data; it does not establish accuracy for people outside the study.
What remains open
- The size and uncertainty of the overall group-preference improvement are not established by the supplied passages.
- The supplied evidence does not establish performance beyond the study data.
Conditions and limits
- This is a reported result from the study’s participant data and validation approach, not evidence that the method works in other settings.
- One supplied result gives numerical gains for 5% and 25% initial samples in a particular augmentation test; it gives no separate numerical result for the 50% sample.
- The answer is attributed to the supplied account of the study, not presented as independent verification.
Evidence and related ideas
Supporting evidence, qualifications and context play different roles in this answer.
supports
Estimates of group preferences Using cross-validation, the authors report that a probability-based sample augmented with model predictions estimated group policy preferences more accurately than the sample alone.
Directly reports that model-augmented probability samples estimated group policy preferences more accurately than samples alone.
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
Testing group-level preferences The researchers also compare whether ordinary random samples and model-augmented samples can reproduce the full group’s combined preferences.
Describes the comparison of ordinary and model-augmented samples but says the supplied passage does not give its outcome.