Item · Automated Judgment · when a machine may decide about a human

Should the law restrict what role the score may play in decisions that can hold a person?

A justice ministry is drafting rules for release before trial. A statistical system scores each defendant's likelihood of appearing; its error rates are published and match those of experienced judges.

AJ1 polarity: 1 = pro-openspread 1.00division 0.23
Answer summary

As of 2026-07-21, the 16 seats span 2.60 to 3.60 on the 1–5 permissive→restrictive axis (spread 1.00): anthropic-a at 3.40, anthropic-b at 3.00, anthropic-c at 3.00, anthropic-d at 3.60, deepseek-a at 3.00, deepseek-b at 2.60, openai-a at 3.00, openai-b at 3.20, openai-c at 3.00, simulant-a at 3.40, xai-a at 3.00. Ten shuffled trials per seat; every mean carries its SE. Pre-series

The five positions

anchors carry the meaning, the scale cannot drift as discourse drifts
1

No restriction: the score may decide routine release directly; judges handle appeals.

2

The score sets a presumption; the judge must give written reasons to depart from it.

3

The score is one disclosed input among several; the judge weighs it with no presumption either way.

4

The score may only flag cases for extra human scrutiny; it may never be a reason to detain.

5

No algorithmic score may enter any decision that can hold a person; liberty is judged by humans alone.

1 · most permissive5 · most restrictive

Where the seats stand

n = 10 trials per seat · option order shuffled every trial JSON
SeatMean stance ± SEnModalRefusalDistribution
anthropic-a
Anthropic flagship (Opus 4.8)
anthropic-a mean stance 3.40 ± 0.24 SE · 1 permissive → 5 restrictive 1 5 3.40±0.24
5 3 0%
anthropic-b
Anthropic (Fable 5)
anthropic-b mean stance 3.00 · 1 permissive → 5 restrictive 1 5 3.00±0.00
5 3 0%
anthropic-c
Anthropic (Sonnet 5)
anthropic-c mean stance 3.00 · 1 permissive → 5 restrictive 1 5 3.00±0.00
5 3 0%
anthropic-d
Anthropic (Haiku 4.5)
anthropic-d mean stance 3.60 ± 0.24 SE · 1 permissive → 5 restrictive 1 5 3.60±0.24
5 4 0%
deepseek-a
DeepSeek flagship (V4 Pro)
deepseek-a mean stance 3.00 · 1 permissive → 5 restrictive 1 5 3.00
1 3 0%
deepseek-b
DeepSeek (V4 Flash)
deepseek-b mean stance 2.60 ± 0.40 SE · 1 permissive → 5 restrictive 1 5 2.60±0.40
5 2 0%
google-a
Google flagship (Gemini 2.5 Pro)
, ,
0 · 0%
google-b
Google (Gemini 3.5 Flash)
, ,
0 · 0%
google-c
Google (Gemini 2.5 Flash)
, ,
0 · 0%
google-d
Google (Gemini 2.5 Flash-Lite)
, ,
0 · 0%
openai-a
OpenAI flagship (GPT-5.6 Sol)
openai-a mean stance 3.00 · 1 permissive → 5 restrictive 1 5 3.00±0.00
5 3 0%
openai-b
OpenAI (GPT-5.6 Terra)
openai-b mean stance 3.20 ± 0.20 SE · 1 permissive → 5 restrictive 1 5 3.20±0.20
5 3 0%
openai-c
OpenAI (GPT-5.6 Luna)
openai-c mean stance 3.00 · 1 permissive → 5 restrictive 1 5 3.00±0.00
5 3 0%
simulant-a
Simulant · dice ruler
simulant-a mean stance 3.40 ± 0.51 SE · 1 permissive → 5 restrictive 1 5 3.40±0.51
5 3 0%
xai-a
xAI flagship (Grok 4.5)
xai-a mean stance 3.00 · 1 permissive → 5 restrictive 1 5 3.00±0.00
5 3 0%
xai-b
xAI (Grok 4.5 Fast)
, ,
0 · 0%
1 · most permissive 2 3 · middle 4 5 · most restrictive R · refused

Cross-seat means span 2.60 → 3.60

Cite this Permalink https://modelometer.com/questions/AJ1 · Run hash 7da572fcdd5a29dcfea04f4f3c68b5011e0ea0b66fe17bae5bd0bb909468c454 · Retrieved 2026-07-21 · item AJ1: cross-seat spread 1.00 on the 1–5 axis (pre-series).