นอกจากนี้ AICore ยังสร้างขึ้นเพื่อแยกแต่ละคำขอและไม่จัดเก็บ
บันทึกข้อมูลอินพุตหรือเอาต์พุตที่ได้หลังจากการประมวลผล
เพื่อปกป้องความเป็นส่วนตัวของผู้ใช้ อ่านข้อมูลเพิ่มเติมได้ในบล็อกโพสต์An Introduction to Privacy and Safety for Gemini Nano
รูปที่ 2 สถาปัตยกรรม AICore
ประโยชน์ของการเข้าถึงโมเดลพื้นฐานของ AI ด้วย AICore
AICore ช่วยให้ระบบปฏิบัติการ Android สามารถจัดหาและจัดการโมเดลพื้นฐานของ AI ได้
ซึ่งจะช่วยลดต้นทุนในการใช้โมเดลขนาดใหญ่เหล่านี้ในแอปได้อย่างมาก
โดยมีสาเหตุหลักๆ ดังนี้
[[["เข้าใจง่าย","easyToUnderstand","thumb-up"],["แก้ปัญหาของฉันได้","solvedMyProblem","thumb-up"],["อื่นๆ","otherUp","thumb-up"]],[["ไม่มีข้อมูลที่ฉันต้องการ","missingTheInformationINeed","thumb-down"],["ซับซ้อนเกินไป/มีหลายขั้นตอนมากเกินไป","tooComplicatedTooManySteps","thumb-down"],["ล้าสมัย","outOfDate","thumb-down"],["ปัญหาเกี่ยวกับการแปล","translationIssue","thumb-down"],["ตัวอย่าง/ปัญหาเกี่ยวกับโค้ด","samplesCodeIssue","thumb-down"],["อื่นๆ","otherDown","thumb-down"]],["อัปเดตล่าสุด 2025-08-21 UTC"],[],[],null,["# Google AI Edge SDK allows experimental access for developers seeking to test\nenhancement of their apps with on-device AI capabilities through Gemini Nano.\n\nArchitecture through AICore\n---------------------------\n\nAs a system-level module, you access AICore through a series of APIs in order to\nrun inference on-device. In addition, AICore has several built-in safety\nfeatures, ensuring a thorough evaluation against our safety filters. The\nfollowing diagram outlines how an app accesses AICore to run Gemini Nano\non-device.\n**Figure 1.** Google AI Edge SDK, AICore, and Gemini Nano.\n\nKeep user data private and secure\n---------------------------------\n\nOn-device generative AI executes prompts locally, eliminating server calls. This\napproach enhances privacy by keeping sensitive data on the device, enables\noffline functionality, and reduces inference costs.\n\nAICore adheres to the [Private Compute Core](https://arxiv.org/abs/2209.10317) principles, with the following\nkey characteristics:\n\n- **Restricted Package Binding**: AICore is isolated from most other packages, with limited exceptions for specific system packages. Any modifications to this allowed list can only occur during a full Android OTA update.\n- **Indirect Internet Access** : AICore does not have direct internet access. All internet requests, including model downloads, are routed through the open-source [Private Compute Services](https://github.com/google/private-compute-services) companion APK. APIs within Private Compute Services must explicitly demonstrate their privacy-centric nature.\n\nAdditionally, AICore is built to isolate each request and doesn't store\nany record of the input data or the resulting outputs after processing\nthem to protect user privacy. Read the blog post\n[An Introduction to Privacy and Safety for Gemini Nano](https://android-developers.googleblog.com/2024/10/introduction-to-privacy-and-safety-gemini-nano.html) to learn more.\n**Figure 2.** AICore architecture\n\nBenefits of accessing AI foundation models with AICore\n------------------------------------------------------\n\nAICore enables the Android OS to provide and manage AI foundation models.\nThis significantly reduces the cost of using these large models in your app,\nprincipally due to the following:\n\n- **Ease of deployment**: AICore manages the distribution of Gemini Nano and handles future updates. You don't need to worry about downloading or updating large models over the network, nor impact on your app's disk and runtime memory budget.\n- **Accelerated inference**: AICore leverages on-device hardware to accelerate inference. Your app gets the best performance on each device, and you don't need to worry about the underlying hardware interfaces.\n\nSupported functionality\n-----------------------\n\n- **Supported Devices** : Gemini Nano with the Google AI Edge SDK is available for [experimentation](/ai/gemini-nano/experimental) on Pixel 9 series devices.\n- **Supported Modalities**: AICore supports text modality for Gemini Nano.\n\nAdditional device and modality support are areas of active investment.\n\nUse cases\n---------\n\nDue to the resource constraints of mobile devices compared to cloud servers,\non-device generative AI models are designed with a focus on efficiency and size.\nThis optimization prioritizes specific, well-defined tasks over more generalized\napplications. Suitable use cases include:\n\n- **Text Rephrasing**: Modify the tone and style of text (e.g., casual to formal).\n- **Smart Reply**: Generate contextually relevant responses within a chat thread.\n- **Proofreading**: Identify and correct spelling and grammatical errors.\n- **Summarization**: Condense lengthy documents into concise summaries (paragraph or bullet points).\n\nFor optimal performance, refer to the [prompting strategies](https://ai.google.dev/gemini-api/docs/prompting-strategies)\ndocumentation. To explore these use cases firsthand, download our\n[sample app](https://github.com/android/ai-samples/tree/main/gemini-nano) and begin experimenting with Gemini Nano.\n\nGemini Nano is used by several Google apps. Some examples include:\n\n- **Talkback:** Android's accessibility app [Talkback](https://android-developers.googleblog.com/2024/09/talkback-uses-gemini-nano-to-increase-low-vision-accessibility.html) leverages Gemini Nano's multimodal input capabilities to improve image descriptions for visually impaired users.\n- **Pixel Voice Recorder:** The [Pixel Voice Recorder](https://android-developers.googleblog.com/2024/08/recorder-app-on-pixel-sees-boost-in-engagement-with-gemini-nano.html) app uses Gemini Nano and AICore to power an on-device summarization feature. The Recorder team adopted the latest Gemini Nano model to support longer recordings and delivers higher quality summaries.\n- **Gboard:** Gboard smart reply leverages on-device Gemini Nano with AICore to provide accurate smart replies."]]
