Making new AI content is now simpler than ever thanks to this experimental capability.
Google launched the Search Generative Experience (SGE) earlier this year in an effort to synthesize search results using AI. Google has added a new twist to its AI search testing that puts an emphasis on visuals rather than text. As of right now, you can make brand-new AI photos by entering a suggestion into the Google search box or the Chrome search bar.
The Google SGE must be turned on via Google Labs because it is an experimental function. The new AI-generated photographs need you to enable this option, and they aren’t always the best. The blue links are hidden by the SGE, which shows up at the top of the search results. The SGE box uses generative AI to try and explain what the links say. It’s useful occasionally, but it happens a lot.
To enable the new picture creation feature, all you have to do is modify your search query. You must instruct Google to “draw” or “make” an image before it will substitute four photos that resemble your prompt for the typical SGE text. Examples of Google’s interpretation of “an abstract image of two technicolor spiral galaxies colliding” are provided below.
At only 512px square, the photos Google returns aren’t particularly large. Yet, the majority of the images’ content is excellent. We created a few hand-related images, which are notoriously challenging for AI image generators, and the majority of them had the right amount of fingers. With your input, Google does appear to take some sweeping liberties. Google placed the corgi we asked to see reading a book on a mountain, in a bathtub, and in space. All of these are acceptable responses, but they are entirely up to the AI’s whims.
There is a different method for obtaining AI photos from Google that may be much more beneficial. You may be able to create an AI image using certain Google image searches. But, this button doesn’t always show up. Google doesn’t go into much detail beyond stating that it “may” occur.
Some have questioned if the enormous cost associated with operating large language models is justified once the initial AI frenzy has subsided. Companies not only have to purchase rack after rack of pricey AI accelerator cards, but the ongoing cost of powering these systems is also enormous. With data centers dotted over the globe and an endless supply of cash, Google is in a better position than most. Even Google won’t carry out these tests indefinitely, though. Either this will shape the future or it will be a passing fad. In either case, starting today, you may view Google’s most recent AI experiments. To get started, simply go to the Laboratories page.