Engineers noted that Google rarely publishes its algorithms. Although the Imagen model produces the result,
AI Test Kitchen app launched early2022 as a way to beta test artificial intelligence systems. The application now offers several ways to interact with the Google LaMDA text model. The company has now added two ways to interact with Imagen.
In City Dreamer mode, users canask the model to generate the elements of the city, decorated according to the theme they have chosen. Imagen creates sample buildings and lots. In this case, all projects are displayed in the form of isometric models, similar to those that can be seen in SimCity.
City dreamer mode
In Wobble mode, you can create a small monster.First you need to choose what it will be made of (clay, felt, marzipan, rubber), and then generate clothes and appearance. The model creates a monster, gives it a name, and then makes it "dance".
These interactions are extremely limited compared towith other text-to-image models, users can't request everything they want. However, this is an intentional limitation on Google's part. As Josh Woodward, Senior Director of Product Management at Google, explained, the whole point of the AI Test Kitchen is to get customer feedback and learn about how users will interact with them.
Wobble Mode
Google did not share data on how muchpeople actually use the AI Test Kitchen. “This is a very active, opinionated group of users,” the company said. “The service is very useful for certain categories of people, for example, researchers.”
Experts unsure what Google will promoteproducts for the general public and beyond. At the same time, the company's competitors, OpenAI and Stability AI, are already commercializing text-to-image models.
Read more:
Archaeologists have officially confirmed the legends from the Bible
The grave of the "priestess" of Aphrodite was found: scientists showed what they found there
Scientists have seen what is on the territory of the Mayan capital. The find surprised them.