Google Search Now Lets You Create Images With AI: Here's How It Works

Google is stepping up its game in the AI-powered image generation arena. Starting this week, users can now generate images directly from text prompts using Google’s Search Generative Experience (SGE). This move comes on the heels of Microsoft‘s integration of OpenAI‘s DALL-E model with Bing Chat earlier this month.

How to create images with Google Search’s generative AI

To use this feature, users who have opted into SGE via Google’s Search Labs program can simply type their desired prompt into the Google search bar. The system then offers a selection of images generated from the prompt, all powered by Google’s Imagen AI models.

As noted in a report by The Verge, a demonstration by Google showcased the tool’s capability to generate images from the prompt “draw an image of a capybara wearing a chef’s hat and cooking breakfast.” Users can also fine-tune the queries to influence the resulting images. Additionally, while browsing Google Images, users might be prompted to generate images based on their search, which will appear in a sidebar.

Google emphasizes the responsible deployment of this tool. Hema Budaraju, in a company blog post, highlighted that generated images will carry metadata and watermarks indicating AI creation. To ensure ethical use, Google has set restrictions, including not generating photorealistic faces or images of notable individuals. The feature is also limited to users aged 18 and above.

But that’s not all. Google is also introducing a draft writing feature within SGE. Users can input a query, and Google will generate a draft, which can be tailored for length or tone. These drafts can then be exported to Google Docs or Gmail. As for the tech behind this, Ewer mentioned that SGE utilizes a mix of LLMs, trained on public data and human supervision.

Read next: Samsung Nears ₹1 Lakh Crore Milestone In India Thanks To Its Smartphone Sales