What Exactly is MCP, Which Robin Li Strongly Promotes? Everyone, Pay Attention!
宝子们，AI 圈又双叒爆火一个新词 ——MCP！百度董事长李彦宏在 Create 大会上说，现在基于 MCP 开发智能体，就像 2010 年开发移动 APP 一样有潜力。这 MCP 和前阵子超火的 Agent（智能体）关系可大了，它是实现 Agent 愿景的重要路径，能让大模型自由调用外部工具，完成各种具体任务。
Guys, there's a new buzzword in the AI circle - MCP! Robin Li, the chairman of Baidu, said at the Create conference that developing agents based on MCP now is as promising as developing mobile apps in 2010. MCP is closely related to Agent, which was super popular recently. It's an important way to realize the vision of Agent, allowing large - language models to freely call external tools and complete various specific tasks.
MCP: The "Super Power - Up" for AI!
MCP 全称是 “Model Context Protocol”，也就是模型上下文协议，是一项开放标准。简单来说，它就像是给大模型装了个 “万能插座”，让基于大模型的应用能和外界数据源、工具互动。以前开发 AI 应用门槛超高，比如做个 AI 旅游助手，开发者得费好大劲让大模型适配各种工具，代码复用性还低。但有了 MCP，开发者不用再为外部工具性能操心，只维护应用就行，开发工作量大大减少。
The full name of MCP is "Model Context Protocol", which is an open standard. Simply put, it's like installing a "universal socket" for large - language models, enabling applications based on them to interact with external data sources and tools. In the past, the threshold for developing AI applications was extremely high. For example, when creating an AI travel assistant, developers had to go to great lengths to make the dydoujia.com large - language model adapt to various tools, and the code reusability was low. However, with MCP, developers no longer need to worry about the performance of external tools and only need to maintain the application, greatly reducing the development workload.
jrhz.info MCP 真有那么神？其实还存在不少问题……Is MCP Really That Great? Actually, There Are Quite a Few Problems...
虽说 MCP 优点多多，但它目前还处于发展初期，远不是完美方案。有些开发者觉得它为建标准而建，API 或许更简洁。而且 MCP 『服务器』如果不是官方的，安全性和稳定性就很难保证。像某度地图的 MCP Server，工具少不说，使用还麻烦，这让开发者很头疼。
Although MCP has many advantages, it's still in its infancy and far from a perfect solution. Some developers think it's just for the sake of establishing a standard, and API might be simpler. Moreover, if the MCP server is not official, its security and stability are hard to guarantee. Take a certain Baidu Map's MCP Server as an example. It not only has few tools but is also troublesome to use, which gives developers a headache.
MCP 爆火背后：开放路线的胜利！Behind the Popularity of MCP: The Victory of the Open - Source Approach!
MCP 刚推出时没多少人关注，后来 AI 编程领域的『明星』️应用纷纷支持，大模型厂商也陆续接入，它才火起来。大厂们拥抱 MCP，其实是生态战略的转向。以前 OpenAI 的生态从开放走向封闭，效果并不好。MCP 更开放、抽象程度更高，让大厂们看到了把蛋糕做大的希望，这也证明了开放、开源路线在前沿科技行业依旧是最优解！
When MCP was first launched, not many people paid attention. It became popular later when star applications in the AI programming field began to support it one after another, and large - language model manufacturers also started to adopt it. The embrace of MCP by big companies is actually a shift in their ecosystem strategy. In the past, OpenAI's ecosystem changed from open to closed, and the results were not satisfactory. MCP is more open and highly abstract, giving big companies hope to expand the market. This also proves that the open - source approach is still the best solution in the cutting - edge technology industry!
