Mastering Race & Ethnicity in ChatGPT o4 Image Generation

Your Practical Guide to More Accurate and Respectful AI Portraits

Why Representation Matters in AI Images

If you've ever tried generating people with ChatGPT's o4 Model and ended up with bland, same-looking faces, you're not alone. Without enough detail, the model tends to default to generic features. But this isn’t a flaw - it’s a lack of clear prompting.

In Prompting People with ChatGPT Image Gen: The Ultimate Reference, I break down exactly how to create more specific and respectful representations of race and ethnicity. This newsletter gives you a glimpse into the "Race & Ethnicity" chapter, showing you how to move beyond stereotypes and get images that truly reflect the diversity you're aiming for.

The Power of Specificity

ChatGPT's o4 Model can generate diverse individuals across nationality, race, gender, facial structure, hairstyle, and more. But page 4 of my book makes one thing clear: without enough detail, you'll get the same default faces every time. The key is guiding the model with clear intent.

Nationality Prompts: A Simple Starting Point

Page 18 explains how just adding a nationality like "Japanese woman" or "Nigerian man" instantly improves the result. But be aware—the model sometimes leans into common stereotypes unless you steer it further. For example, "Irish person" often defaults to red hair unless you specify otherwise.

Pro Tip: Use modern, neutral clothing in your prompts unless traditional attire is part of daily life in that culture.