85.7 / 100
Fit, read from what they wroteMean over 482 of 576 written answers · rubric-v3 · 57 distinct values. Reader-scored — not the residents’ own rating.
Decision
Study
How to read this
Directional decision evidence from a digital society — not observed human behaviour. Percentages are shares of this room, with no margin of error.
A single run supports at most MEDIUM confidence; treat any segment under 48 people as directional.
This ran on the test engine: it validates the study's structure and topline direction. Psychology-level claims come from the deep read.
The decision
Do you think AI does more good than harm, more harm than good, or equal amounts of harm and good? Why?
The clearest answer
Nearly two-thirds of the society came down against AI doing more good than harm, driven by fear that jobs disappear faster than retraining and new work appear.
- A third sat in the middle, uncertain whether the balance tips either way, suggesting the case for AI benefit is unresolved rather than rejected outright.
Stop arguing AI's net benefit.
2 of 576 (0%) could not say what this actually is. Their scores are reactions to something they couldn't identify — read the rest with that in mind.
position: 38.1 — scored by one reader against rubric-v3 across 539 written answers, blind to who wrote them, 61 distinct values. That is the score of record; say it is reader-scored whenever you quote it. The residents' own position ratings did not resolve and stay in the transcript for audit, not in this verdict.
44 / 576
They want the entry price to come down44 of 576 residents asked for it · 8% · directional until tested in the real world
No coherent dissent — the read was broadly shared.
Test whether "They want proof — evidence, results or an example" actually moves 8–12 real people.
This is prework. It tells you where to point a real conversation — not what the answer will be.
The answer is complete above. Open a lens only where deeper inspection changes what you do.
Pattern
Only the signals that change the decision.
The full result remains available, but these are the findings that determine what to do now.
84 of 576 residents asked for it · 15%
single-run theme, n=84 — a second run that repeats it earns HIGH in this society's brief
230 of 576 residents raised it · 40%
single-run theme, n=230 — a second run that repeats it earns HIGH in this society's brief
single-run segment split, n=59 — a second run that repeats it earns HIGH in this society's brief
Read from what 576 people wrote: 63.7% against, 31.6% in the middle, 1.6% for. 18 of 576 were NEVER CLASSIFIED — a classifier batch failed or its reply was cut short, so they sit in no bucket. That is a failure of this engine's read, not people declining to answer. FORCED READ: answers were placed in the closest bucket, as a poll that offers no 'don't know' would. Comparable to such a poll; NOT comparable to this engine's ordinary reads.
Complete family reaction map