The Google Gemini Nano is the smallest multi-language model and it can process generative AI tasks from within a device without Internet access. Gemini Nano uses the AICore Android system application on Pixel 8 and Samsung Galaxy S24 to process transcripts from voice recorders and AI on the Gboard keyboard for Smart Reply and Magic Compose.
Recently, Android Authority found evidence that the Gemini Nano could later process web pages to compile articles. Now the summary of articles on the web is processed in the cloud with the Internet. This is because the code in the Google AI Edge SDK proposes a Protected Download API that can download content from authentic sources to feed to the Gemini Nano. With API LoRA (low-rank adaptation), Gemini Nano can focus on one task which is AI_CORE_CHROME_SUMMARIZATION_OUTPUT.
Same feature by Samsung with Galaxy AI
With Chrome already having AI features to help with writing, Google's search engine with SGE and Microsoft Edge having proven the Chromium web browser can support generative AI, don't be surprised if article summarization features will be built right into Chrome. For now it is still uncertain when Google will roll out this feature to everyone. But it is definitely for premium flagship devices only like Pixel 8 Pro and Samsung Galaxy S24 series.