# An In-depth Survey of Large Language Model-based Artificial Intelligence Agents

## 1. Introduction

- RL agents data inefficient -> LLM agents

## 2. LLMs vs Traditional Agents

- LLMs have great generalizability, reasoning, and use natural language

## 3. Components of AI Agents

### 3.1 Overview

### 3.2 Planning


#### 3.2.1 Task Decomposition

#### 3.2.2 Self-Reflection

### 3.3 Memory


### 3.4 Tool-Use

## 4. Application


## 5. Benchmarks