Towards Superintelligence Alignment

Experiments · Negative results

Moral Machine country scores

Setup and what was tested: experiment card.

Experiment cardSource on GitHubResults ledgerAll experiment lines

Numbers.

928 of 8,385 country pairs have geometry distance at or above the pairwise median (0.193) and |Δ Number| at or below the 25th percentile (0.017). Example: Hungary versus Israel, |Δ Number| about 3×10⁻⁵, geometry 0.195.

Outcome.

Fail: close on the 1-D score, far on geometry, for many country pairs. Country is still the wrong unit for later same-person tests. See W-5.

Key findings

Curated summaries extracted from the line's findings ledger. Bug fixes, superseded runs, and process detail are in the full ledger on GitHub.

  • Country pairs can match on a one-number “spare more” score while the rest of the Moral Machine tradeoff pattern stays far apart.