With Imagen 3 on Gemini, Google hopes to avoid past mistakes in generating people images

Google has recently launched the August 2024 for Gemini, its AI chatbot, with a few additions here and there. With this update, Gemini is finally integrating Imagen 3—its advanced image generation model, just some time after its initial launch.

The tech giant is now reintroducing the ability to generate images of people using the popular AI chatbot, which had been paused earlier this year due to concerns over historically inaccurate images that produce non-white images for historically white figures, among others.

Google also reaffirmed that, unlike xAI’s Grok image generator that lets you push the envelopes (too far, sometimes), it does now allow “the generation of photorealistic, identifiable individuals, depictions of minors or excessively gory, violent or sexual scenes.”

This feature will now be available to Gemini Advanced, Business, and Enterprise users via the upgraded Imagen 3 model, which now includes new safeguards to address previous issues.

“Of course, as with any generative AI tool, not every image Gemini creates will be perfect, but we’ll continue to listen to feedback from early users as we keep improving,” the company promises.

It actually wasn’t a long time ago when Google announced Imagen 3, but a few weeks do feel like an eternity in the AI industry. It was a soft, quiet launch via Google’s AI Test Kitchen as Google is being careful in its approach to the image generator.

Speaking of the August 2024 update, Google is now supporting new file types and adding a custom GPT-like feature called “Gems.” It’s honestly a bit surprising that it hasn’t been a thing already since the Microsoft-backed OpenAI already has it for months, and you can even access its GPT Store.