Opus 4.8 Computes 11.7 Billion Lives and Creates a Human Reincarnation Simulator

Claude Opus 4.8 powered a website called "The Veil of History", inspired by John Rawls' veil of ignorance thought experiment, to translate the fate of every human ever born—approximately 11.7 billion people—into a concrete probability distribution.

To obtain the total number of births, Opus aggregated the 2022 Population Bureau report, the Madison Project database, the HYDE dataset, and United Nations population projections, constructing a matrix of births by era and region and weighting each era by its actual birth count.

The model applied a simple formula—multiplying each era's total births by the region's share of global population and summing across generations—and handled the lack of regional cumulative birth statistics by running a Monte‑Carlo simulation of 4,000 iterations, expanding uncertainty for older, less‑certain periods and reporting median values with 5‑95% confidence intervals.

Opus then built a single‑page, scroll‑driven front‑end using D3 and Natural Earth to render an interactive timeline and a "DRAW A LIFE" button that randomly assigns a birth era, location, socioeconomic status, and lifespan. The visualisation shows that roughly 81 % of all humans were born before 1650 and 94 % before 1900, with the most probable life being that of an illiterate East‑Asian farmer living about 21 years.

The project demonstrates that Opus 4.8 acted as four distinct high‑skill roles—data analyst, mathematical modeler, front‑end architect, and copywriter—delivering a production‑grade site without any human programmer involvement.

In AI benchmark competitions, Opus 4.8 achieved a score of 61.4 on the Artificial Analysis leaderboard, surpassing GPT‑5.5, earned 45.7 % on Humanity's Last Exam, outperformed GPT‑5.5 by 1 point, scored 69.2 % on SWE‑Bench Pro versus 58.6 % for GPT‑5.5, and attained an Elo rating of 1890, roughly 121 points higher than its nearest rival.

These results illustrate that modern LLMs can independently construct complex, data‑driven applications, raising new questions about the boundaries of AI agency and the standards used to evaluate such systems.