Open Spinning Arrow →

What it is

Spinning Arrow is an interactive research project that makes models’ answers to difficult value questions easier to inspect. It compares leading AI systems using established moral-psychology and ethical-decision measures, then turns the results into a clear, browsable map.

Rather than calling any answer good or bad, it asks a simpler question: which values does a model tend to emphasize when the answer involves a real trade-off?

What you can explore

  • Compare models across themes such as care, fairness, loyalty, authority, and purity.
  • See the questions and methods behind each score, in plain language.
  • Drill down from a chart to a model’s exact answer to a particular prompt.
  • Take a short or extended survey to see which tested models have answer patterns most similar to your own.

The project is a transparency and research exercise, not a personality test or a moral scorecard. The goal is to make the value judgments embedded in AI systems more legible—and more discussable.

Explore the full project →