按年被提到
20232023 · 13 次2026
强化学习与智能体›大模型智能体·arXiv:2304.03442 · cs.HC

Generative Agents: Interactive Simulacra of Human Behavior

生成式智能体:人类行为的交互式拟像
Joon Sung Park, Joseph C. O'Brien, Carrie J. Cai, Meredith Ringel Morris, Percy Liang, Michael S. Bernstein · 2023

带记忆、反思与规划的 LLM 智能体在沙盒小镇中模拟人类行为

开始阅读 ↗PDFalphaXiv 讨论
被推荐 · 20 人首次 2023-04-10 · 最近 2026-08-17
20不同的人
14活跃月份
21讨论串
2,650被引(OpenAlex)

论文提出 generative agents,即能模拟可信人类行为的软件智能体,其架构扩展大语言模型,用自然语言存储完整经历记录,随时间把记忆综合成更高层的反思,并动态检索以规划行为。作者在受 The Sims 启发的沙盒小镇中放入 25 个智能体,评估显示它们表现出可信的个体和涌现的社交行为,消融实验表明观察、规划、反思三个组件都对可信度至关重要。

HN 为什么提它

HN 上的人把它当作 LLM 智能体记忆、反思与长期规划的代表作,常在讨论游戏 NPC、多智能体模拟和智能体记忆设计时引用,称它开创了用提示驱动自主 NPC 的思路。

方法难度 2 / 5约 70 分钟经得起时间
读之前最好知道
1大语言模型与提示工程(prompt engineering)
2检索与向量嵌入(embedding)的基本概念
3智能体(agent)架构的基本思想
4人机交互(HCI)评估方法的大意

HN 书评

推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及
强烈推荐localhost · 2023-09

“One of the best papers that I have read on this topic is the Generative Agents paper by Park [1]”

强烈推荐:读过的该主题最好的论文之一
Show HN: LLM Agent Paper List原评论 ↗
推荐akhayam · 2023-04

“I read Park et al.'s paper on Generative Agents a couple of weeks back which seems to have the potential to put LangChain's Memory on steroids: Generative Agents”

推荐 Park 等人的论文,认为它有望大幅增强 LangChain 的 Memory
Autonomous Agents and Agent Simulations原评论 ↗
推荐dan-g · 2023-07

“Generative Agents: Interactive Simulacra of Human Behavior[1]. Make sure to check out the recorded demo!”

推荐这篇论文,并提醒一定要看录制的演示
Ask HN: Which recently research paper blow your mind?原评论 ↗
推荐ykhli · 2023-08

“Inspired by great work from the Stanford Generative Agent paper”

说明自己的项目受斯坦福 Generative Agents 论文的出色工作启发
Show HN: AI-town, run your own custom AI world SIM with JavaScript原评论 ↗
推荐drusepth · 2024-11

“is a really good sneak-peek into the kind of evolving sandboxes LLMs (with memory and decisionmaking) enable.”

推荐:很好地预览了带记忆和决策的 LLM 所能带来的演化沙盒
Project Sid: Many-agent simulations toward AI civilization原评论 ↗
推荐DonHopkins · 2026-01

“For LLM multi-agent work: "Generative Agents: Interactive Simulacra of Human Behavior" (Park et al., Stanford, 2023) Generative Agents”

把它列为 LLM 多智能体方向的推荐论文
Show HN: What if AI agents had Zodiac personalities?原评论 ↗
看全部 20 条 →
引用og_kalu · 2023-04

“Human.exe is a game LLMs can play perfectly. Generative Agents”

引用它说明 LLM 智能体能逼真地模拟人类行为
Emergent abilities of large language models原评论 ↗
引用ianbicking · 2023-05

“Level 6 is reflection as in "Generative Agents: Interactive Simulacra of Human Behavior" Generative Agents”

引用它说明“第 6 级”是反思,即让智能体总结自己的记忆
ChatGPT is Really Good at Roleplaying原评论 ↗
引用og_kalu · 2023-06

“Sure it can. It's not default behavior sure but it's fairly trivial to set up, just expensive.”

回应说模型当然能做到,设置简单只是成本高,并以它作为佐证
The future of AI according to thousands of forecasters原评论 ↗
引用Animats · 2023-07

“Recent progress on that.[1]”

引用它作为 LLM 了解游戏世界方面的近期进展
AI: Startup vs Incumbent Value原评论 ↗
引用advaith08 · 2023-09

“This paper (Interactive Simulacra of Human Behaviour Generative Agents ) has a setup on how we could create a Sims game with us as the characters”

指出论文提供了一种让真人当角色来做 Sims 式游戏的方案
Show HN: WhatsApp-Llama: A clone of yourself from your WhatsApp conversations原评论 ↗
引用araes · 2023-09

“Most text also seems weirdly similar to: Generative Agents: Interactive Simulacra [2]”

指出文章的大部分文字与这篇论文惊人地相似
Things that do not exist in Dimension Apple原评论 ↗
引用dhruvdh · 2023-10

“This makes me think work like "Generative Agents: Interactive Simulacra of Human Behavior" can potentially be run on a single GPU?”

联想到这类工作也许能在单 GPU 上运行
Lamini LLM Finetuning on AMD ROCm: A Technical Recipe原评论 ↗
引用Tantris · 2024-04

“The generative agents paper [1] pioneered the idea of prompting LLMs to create autonomous NPCs.”

指出它开创了用提示让 LLM 驱动自主 NPC 的思路
Show HN: LLM-powered NPCs running on your hardware原评论 ↗
引用drusepth · 2024-08

“Now we're working on NPC simulations based on last year's Generative Agents: Interactive Simulacra of Human Behavior [1] paper.”

说明团队正基于它来做 NPC 模拟
Ask HN: What are you working on (August 2024)?原评论 ↗
引用Philpax · 2024-11

“Much of the base social behaviour here comes from Generative Agents [0], which they cite.”

指出 Project Sid 的基础社交行为很多来自这篇论文,对方也引用了它
Project Sid: Many-agent simulations toward AI civilization原评论 ↗
引用walterfreedom · 2025-11

“I am following some articles loosely, like Generative Agents to take inspiration”

说明自己松散参考它,以获取 LLM 驱动 NPC 行为的灵感
Creating Rimworld Mods to Add LLM PCG and LLM NPCs to Enhance Its RPG Elements原评论 ↗
引用kristianp · 2026-01

“It's kind of surprising that an in game LLM hasn't been done outside of research.”

感叹游戏内 LLM 除了研究之外竟还没人做出来,以它为例
引用ivzak · 2026-03

“Your comment reminded me of this old simulacra paper:)”

想起这篇旧论文,作为压缩智能体记忆的早期例子
玩笑genxy · 2026-08

“If this is too much, maybe you should simulate HN using”

Self hosted email continues to steeply decline原评论 ↗