近日，阿里公布了一项最新的研究成果——手机操作智能体框架Mobile-Agent。该框架具有强大的视觉能力，能跨越APP完成用户交给的任务，实现了即插即用无需训练。此前的手机助手大多局限于单一应用，而Mobile-Agent的推出打破了这一界限，让手机助手真正成为了超级助手。
Mobile-Agent依托多模态大模型，可以玩转10款应用，根据用户需求完成各类任务。例如，当用户需要撰写一篇关于篮球比赛的文章时，只需给出指示，Mobile-Agent便会自动搜索比赛结果，并根据赛况在备忘录中生成文稿。这大大提高了工作效率，让用户得以轻松应对各类任务。
以往，要在不同APP间完成任务，需要编写XML操作文档。而现在，借助Mobile-Agent，这一切都可以通过视觉能力实现，不再需要编写繁琐的文档。这一创新使得APP之间的协同变得更加简单便捷，为用户带来了全新的使用体验。
英文翻译：
News Title: Alibaba Releases Mobile-Agent Intelligent Framework, Breaking App Boundaries and Executing Cross-app Tasks
Keywords: Mobile-Agent, Intelligent Framework, Cross-app Task Execution
News Content:
Recently, Alibaba has announced a latest research achievement – the Mobile-Agent intelligent framework. This framework possesses strong visual capabilities, enabling it to complete user-assigned tasks across different apps, achieving plug-and-play without training. Previous mobile assistants were mostly limited to single apps, but the launch of Mobile-Agent has broken this boundary, making the mobile assistant a true super assistant.
Relying on a multimodal large model, Mobile-Agent can handle up to 10 apps and complete various tasks according to user requirements. For example, when a user needs to write an article about a basketball game, they only need to give instructions, and Mobile-Agent will automatically search for the game results and generate a draft based on the situation. This greatly improves work efficiency and allows users to easily cope with various tasks.
In the past, completing tasks across different apps required writing XML operation documents. Now, with Mobile-Agent, all of this can be achieved through visual capabilities, eliminating the need for tedious documentation. This innovation makes collaboration between apps simpler and more convenient, bringing a new experience to users.
【来源】https://www.qbitai.com/2024/02/118426.html
Views: 1