I Explored Google’s Hidden Open Source App and Discovered the Potential of Offline AI

Google has introduced many AI products recently, with Gemini as its leading virtual assistant. However, AI Edge Gallery, a new app on GitHub, allows users to download large language models (LLMs) to run offline on their phones. It provides a unique opportunity to use these models without internet access, splitting functionalities into categories like Ask Image, Prompt Lab, and AI Chat. While the app’s setup process is complicated—requiring multiple account creations and agreements—its offline capabilities are impressive. For instance, the Gemma model can answer questions about travel planning and explain complex theories, albeit slowly. However, it has limitations; it struggles with specific image recognition and does not function as a full virtual assistant like Gemini. Despite some user experience issues, the app offers potential advantages in privacy and offline usage, making it a worthwhile tool for those with flagship Android devices.

Source link