AI large models have become a hot topic in the current technology circle. More and more companies are beginning to deploy large model capabilities, and more and more products are beginning to emphasize AI. However, judging from current experience, most of the AI products flooding into the market often simply integrate large model applications at the application level, and do not realize systemic AI technology changes from the bottom up.
At HDC2024, with the opening of HarmonyOS NEXT Beta to developers and pioneer users, Huawei showed the industry what true "native intelligence" is - with system-level AI capabilities, AI is no longer just an add-on to mobile phones. Instead, it is deeply integrated with the operating system and becomes a system-level core capability.
According to reports, through the integration of soft and hard core cloud, HarmonyOS NEXT has built a new Hongmeng native intelligence architecture. When AI and OS are deeply integrated, Harmony native intelligence has multi-modal understanding, user personalized data understanding and privacy protection capabilities. Possess full scene perception and collaboration capabilities.
Not only that, in order to make AI better collaborate with and serve people, Huawei and the AIR team of Tsinghua University Intelligent Industry Research Institute jointly released "AI Collaborates with People and Serves People - AI Terminal" white paper". In this white paper, the two parties proposed the concept of the industry jointly building an intelligent classification standard system of L1~L5 - promoting the healthy development of the entire AI terminal industry by formulating industry standards.
Based on its powerful AI base and in-depth understanding of AI terminals, Huawei Xiaoyi is gradually evolving into an intelligent agent, providing users with an unprecedented system-level AI intelligent interaction experience.
After Xiaoyi transformed into an "intelligent agent", the first change brought about was the change of the global entrance and image. Xiaoyi is no longer hidden in The assistant in the background of the system is a visible and tangible existence - Xiaoyi is located on the navigation bar at the bottom of the mobile phone screen. It has the convenient ability to call out the whole world at any time. The user can wake up Xiaoyi by long pressing the navigation bar. At the same time, Xiaoyi also brings a new wake-up animation, a colorful and smart circular three-dimensional design, which adds a strong technological atmosphere and a sense of the future to the interface.
The core meaning of Xiaoyi’s “ubiquity” is to achieve a more intuitive and seamless AI interaction experience.
Taking the AI interaction of traditional mobile phones as an example, users usually need to enter a specific AI function interface first, or call out the voice assistant to state relevant needs. But with Xiaoyi becoming “ubiquitous”, this cumbersome process has been greatly simplified. Users can easily transfer text, pictures, documents and other content directly to Xiaoyi through global drag and drop, realizing instant and intelligent information processing, such as summarization, polishing, table extraction, picture translation, etc.
Take a specific application scenario as an example: For example, when you record a meeting schedule with time and location in a memo, you only need to gently drag this text to Xiaoyi on the bottom navigation bar, and it will Quickly provide you with a schedule pick-up service and automatically add this item to your schedule. For another example, drag and drop a document to Xiaoyi, and it will provide you with multiple functions such as document summary, Q&A, and translation.
Not only that, Xiaoyi also has the ability to understand and identify user intentions, and "see and ask". For example, when receiving a team building email, you can directly ask about the time, distance, surroundings and other details of the address. information and even be able to navigate directly to the location. There is also a streamer reminder function that intelligently recommends related services when copying content, making the AI experience more explicit.
This interaction method is completely in line with the user’s intuitive logic, truly making AI functions within reach in daily operations.
Equipped with Pangu large model, Xiaoyi has strong memory perception capabilities - supports 23 types of TOP scenarios, has stronger reasoning and planning capabilities, and has a task success rate of up to 90%. At the same time, it has mastered trillions of knowledge and achieved seamless connection with 300+ key services.
Xiaoyi has evolved HarmonyOS’s AI system-level intelligent agent, covering Huawei terminal’s 1+8 device product knowledge in multiple dimensions. It can be said to be a treasure trove of Huawei product knowledge in the hands of users, and it can even be called Huawei products. 's encyclopedia.
例如在華為產品相關的專業知識問答場景中，無論是用戶在購買前對設備的詳盡對比、規格查詢，還是購買後對於產品功能有疑問，小藝均能提供詳盡而準確的解答。
舉例來說，當用戶進一步探索某個特定功能，如如何啟用應用分身或掌握隔空手勢的使用技巧時，小藝同樣能夠迅速響應，為用戶帶來全面且專業的指導。未來，小藝也將接觸更多產業知識與應用功能。
在「小藝建議」方面，隨著不斷增強的 AI 引擎技術，小藝透過對使用者行為及高頻使用場景的深度學習和理解，能夠提供更精準的內容推薦和主動連續服務。 HarmonyOS NEXT 下的小藝建議功能，憑藉其與大模型意圖理解能力的緊密結合，不僅深度解析用戶的使用場景，還透過接入更多合作夥伴的服務，為用戶帶來了全新的場景卡體驗。這些智慧推薦服務包括但不限於基於位置的精準推薦，例如當用戶靠近醫院時，小藝會主動推薦取號、電子醫保等便利服務；當用戶接近高鐵站時，則會推薦火車訂票、車次查詢等實用功能。小藝的這些智慧化、個人化的服務，無疑將大幅提升使用者的使用體驗。
在HarmonyOS NEXT 上，華為將AI 能力下沉到OS，構建鴻溝原生智能框架，開啟大模型時代的OS 原生智能新體驗，同時也為開發者提供了生態賦能的卓越特性與能力。例如意圖框架升級，為開發者提供了將應用程式、服務、知識、數據等資源無縫連接到小藝生態的機會。透過 HarmonyOS 系統的原生能力，開發者透過簡單的配置和上傳，便能快速建構基於 AI 模型的智能體。
華為還將AI 能力融入HarmonyOS NEXT 的子系統內，構建了原生智能底座，實現了系統AI 能力與控件的深度融合，為開發者提供AI 賦能的系統控件，降低開發者使用AI 的門檻。
同時，AI 全面賦能OS 子系統，使得華為自有應用的體驗得到了顯著提升，例如瀏覽器朗讀控制、華為帳號實名認證活體偵測圖庫圖文翻譯/ 多主體辨識、智慧搜尋自然語言搜尋等控制功能等，大幅提升華為自有應用的使用體驗。
此外，HarmonyOS NEXT 針對基礎、高頻和創新場景，開放控件，控件AI 化後，開發者能夠「零成本」實現應用的AI 化，用戶也能享受到跨應用一致的AI 交互體驗，這一系列升級，進一步推動了HarmonyOS 向原生智慧型作業系統的邁進。
總的來說，HarmonyOS NEXT 以其前瞻性的理念，不僅為生態系統注入了活力，為開發者提供了強大的支持，更為用戶帶來了前所未有的便捷服務。
鴻蒙不僅是一個作業系統，更透過將 AI 能力下沉至系統層構築了一個原生智慧底座，為我們展現了鴻蒙生態的新可能。隨著 HarmonyOS NEXT 開放給開發者和先鋒使用者 Beta，未來關於「AI」與「OS」組合的前景也越發明朗，或許預示著全場景智慧體驗新時代的來臨。
以上是HarmonyOS NEXT 原生智慧開啟 AI 大模型時代的 OS 新體驗！華為小藝變身智能體的詳細內容。更多資訊請關注PHP中文網其他相關文章！
