Contribution to the handbook
Supporting · Assessed 2026-10-28
The evidence helps qualify an AI-versus-human preference comparison: AI statements were longer on average, length affected ratings and rankings, and the endorsement preference remained after controlling for length. This is useful scrutiny of an alternative explanation, but the supplied records do not establish a governance mechanism or its performance beyond the study.
Does this challenge the goal?
None identified
The question tests an alternative explanation for a reported preference; it does not contest the handbook’s goals or boundaries.
The reported length adjustment is limited to the study’s statement ratings and does not settle other explanations or transfer to public decisions.
Why this category?
- Useful check of a competing explanation for a reported outcome
- Evidence narrows, but does not eliminate, the role of length
The records share a source root. They do not show that length had no effect on other ratings or rankings, or that the same comparison applies to real public deliberation.
This is an AI assessment of contribution to the goal, separate from whether the question has an answer.
Tests, stop conditions and unresolved work →
Read the exploration goalCompare statement lengths and explain how length related to ratings and rankings, including whether the endorsement difference persisted after controlling for length; do not claim length had no effect.
What we can say so far
AI-written statements were longer on average: 125.9 words (95% CI 121.4–130.1), compared with 105 words (100.2–108.5) for human-written statements. Length affected participant ratings and rankings, so it may have contributed to the observed preferences. However, the endorsement difference persisted after researchers controlled for length; length therefore did not account for that difference entirely.
What remains open
- The supplied passage does not quantify how much length affected ratings or rankings beyond reporting that it did.
Conditions and limits
- The supplied evidence is an AI interpretation of the article, not independent new evidence.
- The source passage says length affected ratings and rankings but does not specify the size or direction of those effects.
Evidence and related ideas
Supporting evidence, qualifications and context play different roles in this answer.
context
People rated AI statements more highly Participants preferred AI-generated group statements to those written by human mediators. They rated the AI statements as more unbiased, while external judges rated them higher for quality, clarity, informativeness, and perceived fairness.
Provides general findings on participant and external-judge ratings, but does not address statement length.
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
Participants favored AI statements In 75 six-person testing groups, participants preferred the AI helper’s statements over those written by trained human mediators in 56% of comparisons, and rated the AI statements as more strongly endorsed and higher in argument quality.
Provides preference and endorsement context, but does not address statement length or its control.
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
AI statements were longer on average AI-written statements averaged 125.9 words (95% confidence interval 121.4–130.1), compared with 105 words for human-written statements (100.2–108.5). Length affected ratings and rankings. However, participants still preferred AI statements on endorsement ratings after the researchers controlled for length.
Directly reports the word-count comparison, length’s effects on ratings and rankings, and the endorsement difference after controlling for length.