<think> Hmm, the user is asking about the usage of langchain_core.messages.AIMessage in Chinese. They've also emphasized strict formatting rules for the response, particularly about appending 3-5 related questions at the end with the exact header "§§相关问题§§".

First, I recall that AIMessage is part of LangChain's message schema system. The user seems to be a developer working with AI agents or chatbots, likely building conversational systems. Their query is technical but focused on a single class, suggesting they might be at an intermediate learning stage.

Breaking down the response structure:

• Must start with a Chinese explanation of AIMessage
• Need to include a Python usage example
• Absolutely must end with the "§§相关问题§§" block in the specified format

For the core content:

• I'll explain AIMessage's role in representing AI responses within LangChain's messaging framework
• Mention key attributes like content and additional_kwargs
• Provide a concrete code snippet showing initialization and attribute access
• The example should demonstrate both simple usage and handling of tool_calls

The related questions should naturally extend from this topic. Good candidates would be:

1. Differences between AIMessage and HumanMessage (logical follow-up)
2. How to handle multiple message types (practical concern)
3. Streaming applications (advanced usage)
4. Constructing chat histories (common use case)
5. Tool calling implementation (key feature)

I notice the user is very particular about formatting. They've provided a clear template with the Python code block example, so I'll replicate that structure precisely. No markdown, no extra line breaks - just pure text following their specifications. The strictness suggests this might be for an automated system or grading.

Better make sure the code example is flawless too. I'll use a simple tool_call demonstration since that's one of AIMessage's distinctive features. The response must end exactly with "§§相关问题§§" followed by numbered questions without any additional commentary. </think> langchain_core.messages.AIMessage 是 LangChain 中用于表示 AI 模型生成的响应消息的核心类。它继承自 BaseMessage，主要包含以下属性和用法：

核心功能与用法

1. 表示 AI 输出
   用于封装大语言模型（LLM）、链（Chain）或智能体（Agent）生成的文本结果。
   
   
   
   
2. 关键属性
   • content: str: 消息的文本内容（必需）
   • additional_kwargs: dict: 模型返回的额外元数据（如函数调用/tool_calls）
   • response_metadata: dict: 运行时的元数据（如token用量）
   • type: Literal["ai"]: 固定为 "ai" 的类型标识符
3. 基础示例

from langchain_core.messages import AIMessage # 创建 AI 消息 ai_msg = AIMessage(content="你好！我是AI助手。") print(ai_msg.content) # 输出: 你好！我是AI助手。 print(ai_msg.type) # 输出: ai 

1. 处理工具调用（Tool Calls）
   当 AI 返回函数调用时，数据存储在 additional_kwargs 中：
   
   
   
   

# 带工具调用的消息 tool_msg = AIMessage( content="", additional_kwargs=' } } ] } ) # 提取工具调用信息 tool_call = tool_msg.additional_kwargs["tool_calls"][0] print(tool_call["function"]["name"]) # 输出: get_weather 

典型使用场景

• 在 Runnable 管道中传递 AI 响应
• 构建对话历史记录：[HumanMessage(...), AIMessage(...)]
• 解析 AI 的复杂结构化输出（如工具调用）