Google’s New App: AI on Your Phone, Offline Image Generation & Code Creation

In a quiet but significant move last week, Google rolled out a new application designed to bring the power of open-source AI directly to your smartphone. The app, dubbed the Google AI Edge Gallery, allows users to run various AI models sourced from the burgeoning AI development platform, Hugging Face.

While cloud-based AI models often boast superior processing power, they come with their own set of caveats. Concerns surrounding data privacy and the persistent need for an internet connection have prompted a shift toward on-device AI. The AI Edge Gallery addresses these issues by offering a way to leverage AI even without Wi-Fi or a cellular network.

Described by Google as an “experimental Alpha version,” the application is readily available for download and installation via GitHub. The user interface provides quick access to key AI functionalities, including “Image Question Answering” and “AI Chat.” Upon selecting a function, users are presented with a roster of AI models tailored to the task, such as Google’s own Gemma 3n.

Further showcasing its practicality, the Google AI Edge Gallery includes a “Prompt Lab.” This feature enables users to execute one-off tasks, like summarizing or rewriting text, using the select AI models. The Lab is packed with task templates and adjustable parameters, giving users the flexibility to fine-tune the model’s output and tailor it to their specific needs – a boon for anyone looking to utilize AI for productivity gains.

Google cautions that performance will vary depending on the user’s device. While modern devices with more robust hardware will naturally experience quicker processing speeds, the size of the AI model in question also plays a critical role. Larger models, particularly those tackling complex tasks like image-based queries, will invariably take longer to execute than their more compact counterparts.

In a nod to its community-driven ethos, Google is encouraging developers to offer feedback on their experiences with the Google AI Edge Gallery. The application’s licensing under Apache 2.0 allows for free and unrestricted use, both commercial and otherwise, reflecting Google’s commitment to fostering open-source AI innovation.