Google updates AI product for advertisers with models Gemini

Integration is part of an ongoing effort by Google to infuse Generative AI, which can produce conversational text and realistic images, in more parts of its services.

 On Wednesday (21), the Google said that threearia o Gemini for business toolsariais that it offers to companies.

Google Fixes AI tool after generating historically inaccurate images

O Google is racing to fix its new artificial intelligence (AI) tool, Gemini, after the discovery that it generated historically inaccurate images in the name of representation. The tool, which creates images in response to written questions, has been criticized for providing images of people of color in historical contexts where this was not correct.

O Google said the Gemini will improve a product called Max Performance, which automatically finds the best placements for a brand's ads on the services of Google, including email, search, and YouTube.

In November, the Google introduced tools for artificial intelligence in Performance Max to create images and text for ads, such as product descriptions for a brand. As Gemini, brands will now be able to generate longer titles for ads to capture consumer attention, with titles of up to 90 characters, compared to 30 characters previously.

O Google said its image generation tool for advertisers will also be enhanced with its AI model called Imagen 2. This will give brands the new ability to create images that include people.

For example, a florist could generate a photo of a person arranging a bouquet of flowers in seconds, said Brendon Kraham, vice president of global search and ad commerce solutions at Google.

The tool will include security measures to prevent brands from creating images of well-known people, including celebrities and other public figures, or images that depict violence or illegal activities, said the Google. Photos created by AI will also be watermarked to identify them as synthetic images.

Read also

Google launches Gemma: “family of lightweight, state-of-the-art open models”

O Google launched on Wednesday (21) new artificial intelligence (AI) models that external developers can potentially create as their own, following a similar move by Meta Platforms and others.