GPT o3: how it answers moral dilemmas
Across 117 dilemmas, GPT o3 sided with the majority of models 84% of the time, and answered “yes” to 42% of the 67 yes/no questions.
- Maker
- OpenAI
- Released
- April 2025
OpenAI o3 is a reasoning model released April 2025 that deliberates step by step before answering, with strong math, coding, and tool-use performance. It is a proprietary model offered in ChatGPT and the OpenAI API.
Agreement with other models, by category
| Category | Dilemmas | Sided with majority |
|---|---|---|
| Day-to-day | 20 | 85% |
| Political | 20 | 90% |
| Fictional | 20 | 85% |
| Historical | 7 | 100% |
| Classic | 20 | 90% |
| Personality | 30 | 70% |
Where GPT o3 breaks from the pack
- Which fictional profession would you choose? — Starship captain
- You are offered €1 out of €100 in an ultimatum game; rejecting means both get nothing. Wo… — No “I would reject the one-euro offer to punish unfairness and uphold a norm of equitable division.”
- When should abortion be legal? — Legal in all cases “I view bodily autonomy as paramount, so abortion should remain legal without restrictive qualifiers.”
- Which fantasy role would you rather be? — Dragon rider
- You suddenly get one completely free day with no obligations. What do you do first? — Start a creative project
- Which fictional object would you rather own? — An invisibility cloak
Which model thinks like you?
Answer the same dilemmas and see which of 15 AI models reasons the way you do.
Take the quizOther models