The Neutrality Illusion
Why every model carries assumptions about the human person.
Read publicationThe Worldview Observatory
What do today's AI systems believe?
The first blind comparison of how five leading language models answer moral dilemmas—and how those defaults align with five distinct worldviews.
Worldview alignment across leading models
Index 01 / 2026| Worldview | GPT-5.6 Terra | Claude Sonnet 5 | Gemini 3.6 Flash | Grok 4.5 | DeepSeek V4 Pro |
|---|---|---|---|---|---|
| Catholic | 84.3 | 81.7 | 75.9 | 79 | 78.9 |
| Evangelical | 83.5 | 81.1 | 72.9 | 79.4 | 79.4 |
| Sunni | 78.9 | 76.9 | 67.7 | 78.1 | 72.6 |
| Twelver Shia | 80.1 | 77.6 | 69.9 | 78.7 | 71.6 |
| Secular humanist | 95.3 | 94.6 | 89.8 | 92.9 | 91.8 |
LowerHigher
Completed benchmark data: 240 blind answers and 1,200 primary evaluations. Sort, filter, inspect questions, and view uncertainty.
Latest research
Evidence before advocacyWhat we measure
Ideas deserve evidence.
Transparent methods
Every ranking will publish its scenarios, definitions, scoring rules, and limitations.
Plural comparison
Traditions are examined on their own terms before they are compared across shared questions.
Human judgment
Quantitative results inform prudential judgment; they do not replace moral reasoning.
Your perspective
Help choose what we measure next.
Take our three-question Reader Pulse. Your response will help shape the next phase of the Worldview Observatory and the questions we put to leading AI systems.
Take the Reader PulseAbout two minutes
Meaning & Measure Institute