• Source:JND

AS TOP tech companies continue to stay relevant in this year of evolving AI, users are seeing a bunch of new features and tools. Users are already scrutinising chatbots in search engines and producing images by entering commands for AI models.

Given these recent developments, Google is working to bundle Imagen text-to-image generator in its Gboard for Android, as per 9to5Google.

The strings in the beta version 12.7.05.507749191 have a reference to the AI image generator. It will be available in the "shortcuts strip/page" similar to how clipboard and one-handed mode icons appear in Gboard, hints the report. This has added fuel to the fire of speculation that Imagen might soon debut in Gboard.

Unveiled in May 2022, Google's Imagen text-to-image generator was launched amid the buzz of OpenAI's DALL-E 2. Imagen is advertised for its "Unprecedented photorealism" and "deep level of language understanding." It is a text-to-image diffusion model.

To compare the image generator models, "DrawBench, a comprehensive and challenging benchmark for text-to-image models," was deployed by Google. Post comparing Imagen with VQ-GAN+CLIP, Latent Diffusion Models, and DALL-E 2, the company wrote "human raters prefer Imagen over other models in side-by-side comparisons, both in terms of sample quality and image-text alignment."

The integration will allow users to be creative alongside expressive with the images they want the AI to make for them. This will help users of the keyboard generate images with impressive realism while communicating with their peers in social circles. However, this will have a whole new set of limitations.

Google also announced Bard as a competition to OpenAI's ChatGPT which is undergoing testing after it delivered incorrect answers in the initial demo. After the affair, Pichai asked employees to devote time from their working hours exclusively to test and improve the AI service. The tech giant has more than 20 AI products in the pipeline, according to New York Times.

Also In News