Google’s text-to-image AI model Imagen makes public appearances

Google is extremely cautious about releasing its text-to-image AI system, although the company’s Imagen model produces output quality comparable to OpenAI ‘s DALL-E 2 or Stability AI’s Stable Diffusion, Google has not previously opened the system to the public.

Today, though, the search giant announced that it will be adding Imagen to its AI Test Kitchen app in a very limited form as a way to gather early feedback on the technology.

Join us on Telegram

The AI ​​Test Kitchen was launched earlier this year to test Google’s various AI systems. Currently, the app offers a few different ways to interact with Google’s text model LaMDA, and the company will soon add similarly restrictive Imagen requests as part of what it calls a “second season” update to the app. In short, there will be two ways to interact with Imagen: Urban Dreamer and Wobble.

In Urban Dreamer, users can ask the model to generate urban elements designed around a theme of their choice, e.g. pumpkin, or denim. Imagen created sample buildings and parcels (city plazas, apartment buildings, airports, etc.), all designed in isometric models similar to those seen in SimCity.

In Wobble, you can create a little monster. You can choose its material (clay, felt, marzipan, rubber) and dress it in the clothes of your choice. The model will spawn your monster, give it a name, and then you can poke it and make it “dance”.

Moreover, these interactions are very limited compared to other text-to-image modes, and users can’t just create whatever they like. However, this is also what Google intends to do. Josh Woodward, senior director of product management at Google, explained that the whole point of the AI ​​Test Kitchen is to get public feedback on these AI systems to test which behaviors would make the system crash.

The big question, though, is whether Google will want to bring these models to the wider public, and what form will that take? At present, the company’s competitors OpenAI and Stability AI are eager to commercialize text-image models. Will Google feel that its system is safe enough to go out of the AI ​​Test Kitchen and provide it to users?

(via)