DALL-E 3’s evolution indicates a careful fine-tuning of this generative AI model. While we’ve seen similar changes in the past, such as Midjourney’s meticulous training, the changes occurring within DALL-E are noticeably more profound.
It shies away from generating content associated with gender, and at the mention of “blood,” it attempts a creative transformation into vibrant rainbows. Any hint of violence triggers a complete redraw of the image, suggesting a deep analysis of its output. The model even appears to replace adult content with depictions of ancient art, all while maintaining a censorship-like approach.
Comments