Google’s Pixel Studio now generates images of people, but with unsettling flaws

Google‘s Pixel Studio AI image generator has finally added the ability to create images of people, but the technology shows significant limitations in its early implementation. The update to the Pixel 9’s exclusive image generation app represents Google’s step toward competing with more advanced AI image generators like Midjourney and Apple‘s Image Playground, though the inconsistent quality of its human representations—particularly when generating certain professions—highlights the ongoing challenges in AI-generated human imagery.

The key update: Google has expanded Pixel Studio to generate images of people, a capability previously absent from the AI image generation tool exclusive to Pixel 9 phones.

• The feature began rolling out recently after being announced a couple of weeks ago, marking a significant expansion of the app’s creative capabilities.
• Google acknowledges the feature is still being “fine-tuned,” warning users they might encounter unexpected or blocked results.

Early results reveal limitations: Testing shows inconsistent and sometimes disturbing outputs when generating images of people in certain roles.

• One attempt to generate “an athlete” produced a shockingly inappropriate image of a person without skin, demonstrating significant issues with the implementation.
• The system appears to show gender bias, defaulting to male representations when prompted to create images of professionals like doctors, paramedics, and police officers.

How it compares: Pixel Studio offers more realistic image generation capabilities compared to some competitors but with varying quality.

• Unlike Apple’s Image Playground, which primarily produces cartoon-like images, Pixel Studio can generate photo-realistic humans similar to advanced AI platforms like Midjourney.
• The app supports various artistic styles beyond realism, including cartoon and watercolor renditions when specifically requested.

The bigger picture: AI image generation is advancing rapidly across multiple platforms and services.

• Google recently added native image generation to its Gemini AI, though it’s currently limited to the Google AI Studio developer tool.
• Competitors have also expanded their capabilities, with ChatGPT adding native image generation to its 4o model and Grok 3 introducing image editing features.