To counter Microsoft, Google is testing Search Generative Experience, a tool powered by generative AI accessible directly in the search engine field. A way to make your tool accessible to everyone!
Google continues its frantic race in AI against its great rival Microsoft, which seems to have taken a lead by deploying its technology in a large number of its services, and is gradually catching up. While the Redmond firm has been making it possible to create AI-generated images in the Bing Conversation sidebar since last spring, Google announced, in a blog post dated October 12, 2023, introduce the ability to create images with Search Generative Experience (SGE), an AI-powered tool, accessible directly in the Google search bar. So, if we write “draw a capybara wearing a chef’s hat and preparing breakfast”, the search engine will offer, like its counterpart in Edge, four images generated by AI. There is also an option to create new content in Google Images. “This feature is designed to appear when you search for inspiration, for example ‘minimalist tables for Halloween’ or ‘spooky dog house ideas’”, explains the company. It’s very useful “when you are looking for a specific image, but you can’t find exactly what you have in mind”. But be careful not to overdo it!
Search Generative Experience: image generation within everyone’s reach
By opening one of the images generated in the side panel, we can see that the AI has supplemented the initial query with descriptive details allowing us to arrive at the result in question, such as “a photorealistic image of a capybara wearing a chef’s hat and preparing breakfast in a forest, grilling bacon”. By clicking on the image, you can then modify the query to refine the rendering and obtain exactly what you want. For example, we might prefer to have the capybara cook hash browns instead of bacon, or add a light blue background with clouds. You can then export the result and choose to save the final image to Google Drive, download it to your device or copy it. Google says there will be daily limits on image generation, but it didn’t provide further details.
Of course, we must set limits and develop safeguards with this type of AI, in order to prevent it from being used to generate inappropriate or illicit content – violent, sexual, child pornography, or even deepfakes. of public figures – and Google is perfectly aware of this. Also, the tool is reserved for Internet users over 18 years old, cannot generate photorealistic images of famous people and refuses to create content containing inappropriate terms. In addition to all this, each photo is tagged with an in-house marker, called SynthID, capable of identifying AI-generated content through metadata. Moreover, a tool called “About This Image” will help users assess the context and credibility of images. For example, it will indicate when a similar version of this image was first seen by Google, and show other web pages using a similar image, including news or fact-checking sites.
At the same time, the Search Generative Experience (SGE) adds text generation functions, like what Bard offers. We can therefore have him write notes and other drafts. Customization options include the ability to shorten or lengthen the text and change the tone to be less formal. Once the text is finished, you can export it to Google Drive or Gmail. However, the tool is not yet available to the general public. To use it, you must be registered with Search Labs, a Google experimentation laboratory only accessible from the United States and on a waiting list. And it’s impossible to cheat using a VPN, because the account must have been created specifically from the United States.