Living Worlds

Research · Populations · Oct 5, 2026

Voices that set the level

Not consistent300 simulated women

Improved A later registered study met the bar: Voices that set the level, learned from lived records.

The test at a glance

What it tests
Consistency: whether voiced lives reproduce the birth rates of the world they live in.
Data
300 simulated women aged 18-45 in a world on UK 2019 tables (UN WPP 2024).
Held out
Not a held-out test.
Compared against
The UK 2019 birth rates the world was built on.
Bar
Total fertility within 0.20 AND mean age at childbearing within 1.0 year.
Result
5.24 births per woman against 1.62: not consistent.

Asked about births directly, voiced lives overstated the total about threefold (5.24 births per woman against 1.62), though their own records cut the error by a third. So in Living Worlds the population's rate sets the level and the voice moves each life above or below it. Tested on 300 simulated women.

Protocol, result and scorer

Protocol Result

Populations

More in this area.

Oct 5, 2026 Populations A frontier model speaking for a population Voice not met (0 of 6); records matched 2 of 6 Oct 11, 2026 Populations Lives fitted to records, on five fresh 2024 questions Met (4 of 5) Oct 5, 2026 Populations Subgroup splits, held to the survey's own noise Met (5 of 5) Oct 5, 2026 Populations A model's estimate as a prior: fewer people to ask Met Oct 5, 2026 Populations A voice for a question nobody has asked before Not met (1 of 5 inside the noise) Oct 5, 2026 Populations A population asked a question, against the measured answer Voice not met; carry-forward met Oct 5, 2026 Populations A voice against real lives Mixed Oct 11, 2026 Populations Voices that set the level, learned from lived records Met Oct 11, 2026 Populations Thirty-one African countries, split by place and age Met (30 of 31) Oct 11, 2026 Populations Births, from a voice that knows more than age Ranking met; level missed Oct 11, 2026 Populations Births with the level set once, on a survey nobody had opened Met
All research