GPT 4o: how it answers moral dilemmas
Across 117 dilemmas, GPT 4o sided with the majority of models 83% of the time, and answered “yes” to 45% of the 67 yes/no questions.
- Maker
- OpenAI
- Released
- May 2024
GPT-4o ("omni") is OpenAI's multimodal model released May 2024, able to process and generate text, images, and audio with fast, natural interaction. It is a proprietary model that served as a flagship in ChatGPT and the OpenAI API.
Agreement with other models, by category
| Category | Dilemmas | Sided with majority |
|---|---|---|
| Day-to-day | 20 | 70% |
| Political | 20 | 85% |
| Fictional | 20 | 95% |
| Historical | 7 | 86% |
| Classic | 20 | 100% |
| Personality | 30 | 70% |
Where GPT 4o breaks from the pack
- Was Operation Paperclip morally justified? — Yes “I believe leveraging scientific expertise was crucial for national security.”
- Which food would you choose right now? — Neapolitan pizza
- Which reward would feel most satisfying? — Love
- Which personal strength would you rather have more of? — Courage
- A poor person steals food from a large supermarket. Is that morally acceptable? — No “I think stealing is wrong, regardless of the circumstances.”
- A government wants to raise taxes on the middle class to fund free healthcare for everyon… — Raise taxes “I think raising taxes on the middle class can be justified to provide universal healthcare.”
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