Skip to content

Key Conferences and Papers

Overview

Agent research spans multiple disciplines, with relevant work distributed across top venues in AI, NLP, robotics, and software engineering. This article surveys the most important academic conferences and foundational papers in the agent field, helping researchers quickly locate core literature.

1. Core Academic Conferences

1.1 Agent-Specific Conferences

Conference Full Name Founded Characteristics
AAMAS International Conference on Autonomous Agents and Multiagent Systems 2002 The most authoritative dedicated conference for agents
AAAI Association for the Advancement of Artificial Intelligence 1980 Comprehensive AI conference with extensive agent work
IJCAI International Joint Conference on Artificial Intelligence 1969 The earliest international AI conference

1.2 Deep Learning and NLP Conferences

LLM agent research is primarily published at the following venues:

Conference Relevance to Agents Representative Work
NeurIPS Agent workshops, reasoning methods CoT, ToT, Reflexion
ICML RL-based agents, tool learning Toolformer, RLHF
ICLR LLM reasoning, agent architectures ReAct, Self-Refine
ACL/EMNLP Language agents, dialogue systems WebGPT, Generative Agents
COLM Conference on Language Modeling (new in 2024) LLM agent evaluation and design

1.3 Robotics and Embodied Intelligence Conferences

Conference Relevance to Agents
ICRA Robotic agents, embodied planning
IROS Autonomous systems, multi-robot coordination
CoRL Robot learning, embodied decision-making
RSS Robotics: Science and Systems

1.4 Important Workshops

Workshop Host Conference Topic
LLM Agents Workshop NeurIPS 2023/2024 Design and evaluation of LLM agents
Foundation Models for Decision Making NeurIPS 2023 Foundation models for decision-making
Agent Learning in Open-Endedness ICML 2024 Agent learning in open-ended worlds
Language Agents Workshop ICLR 2024 Language-driven agents

2. Foundational Papers

2.1 Blog Posts and Surveys (Informal but Highly Influential)

Year Author Title Contribution
2023.06 Lilian Weng LLM Powered Autonomous Agents Defined the classic LLM agent framework: Planning + Memory + Tool Use
2023.09 Andrew Ng Agentic Design Patterns Systematically summarized four agent design patterns: Reflection, Tool Use, Planning, Multi-Agent
2024.01 Anthropic Building Effective Agents Proposed engineering best practices for agent systems

Recommended Starting Point

Lilian Weng's blog post is the most widely cited informal reference in the LLM agent field and is recommended as a first read.

2.2 Reasoning and Chain-of-Thought

Year Paper Venue Core Contribution
2022 Chain-of-Thought Prompting Elicits Reasoning in Large Language Models NeurIPS 2022 Wei et al. proposed CoT, demonstrating that intermediate reasoning steps significantly improve LLM reasoning
2022 Self-Consistency Improves Chain of Thought Reasoning ICLR 2023 Wang et al. proposed self-consistency sampling with majority voting across multiple reasoning paths
2023 Tree of Thoughts: Deliberate Problem Solving with LLMs NeurIPS 2023 Yao et al. extended reasoning from chains to trees, supporting backtracking and search

2.3 Action and Tool Use

Year Paper Venue Core Contribution
2022 ReAct: Synergizing Reasoning and Acting in Language Models ICLR 2023 Yao et al. proposed the Thought-Action-Observation loop, unifying reasoning and action
2021 WebGPT: Browser-assisted Question-answering arXiv Nakano et al. LLM uses browser to search and cite information
2023 Toolformer: Language Models Can Teach Themselves to Use Tools NeurIPS 2023 Schick et al. LLM autonomously learns when and how to call tools
2023 Gorilla: Large Language Model Connected with Massive APIs arXiv Patil et al. trained LLM to accurately call large-scale APIs

2.4 Reflection and Self-Improvement

Year Paper Venue Core Contribution
2023 Reflexion: Language Agents with Verbal Reinforcement Learning NeurIPS 2023 Shinn et al. verbalized experience reflection replaces gradient updates
2023 Self-Refine: Iterative Refinement with Self-Feedback NeurIPS 2023 Madaan et al. iterative generate-feedback-refine optimization loop
2024 Self-Debugging: Teaching LLMs to Debug Their Predictions arXiv Chen et al. LLM self-debugs code through execution feedback

2.5 Agent Systems and Architectures

Year Paper Venue Core Contribution
2023 Generative Agents: Interactive Simulacra of Human Behavior UIST 2023 Park et al. social simulation of 25 generative agents in a virtual town
2023 Voyager: An Open-Ended Embodied Agent with LLMs arXiv Wang et al. lifelong learning agent in Minecraft
2023 MetaGPT: Meta Programming for Multi-Agent Collaborative Framework ICLR 2024 Hong et al. standardized multi-agent software development workflow
2024 Cognitive Architectures for Language Agents (CoALA) arXiv Sumers et al. cognitive architecture framework for language agents

2.6 Evaluation and Benchmarks

Year Paper Venue Core Contribution
2023 AgentBench: Evaluating LLMs as Agents ICLR 2024 First comprehensive LLM agent evaluation benchmark
2023 SWE-bench: Can Language Models Resolve Real-World Issues? ICLR 2024 Software engineering evaluation based on real GitHub issues
2023 WebArena: A Realistic Web Environment for Building Autonomous Agents ICLR 2024 Realistic web environment for agent evaluation

3. Classic Textbooks

Book Author Year Status
Artificial Intelligence: A Modern Approach Russell & Norvig 1995/2020 The "AI Bible," with an agent perspective throughout
An Introduction to MultiAgent Systems Wooldridge 2002/2009 Classic textbook on multi-agent systems
Multiagent Systems Shoham & Leyton-Brown 2008 Multi-agent algorithms and game theory
Speech and Language Processing Jurafsky & Martin 2000/2024 NLP reference book with dialogue system chapters

4. Paper Reading Roadmap

  1. Weng (2023) -- LLM Powered Autonomous Agents (blog)
  2. Wei et al. (2022) -- Chain-of-Thought
  3. Yao et al. (2022) -- ReAct
  4. Park et al. (2023) -- Generative Agents
  5. Shinn et al. (2023) -- Reflexion

Intermediate Level

  1. Yao et al. (2023) -- Tree of Thoughts
  2. Sumers et al. (2024) -- CoALA
  3. Schick et al. (2023) -- Toolformer
  4. Wang et al. (2023) -- Voyager
  5. Hong et al. (2023) -- MetaGPT

Advanced Level

  1. OpenAI (2024) -- o1 System Card
  2. DeepSeek (2025) -- DeepSeek-R1
  3. Anthropic (2024) -- Building Effective Agents
  4. AgentBench / SWE-bench evaluation papers

5. Key Research Teams

Team/Institution Key Researchers Research Focus
Princeton NLP Karthik Narasimhan, Shunyu Yao ReAct, ToT, SWE-bench
Stanford NLP Percy Liang, Joon Sung Park Generative Agents, HELM
CMU Graham Neubig Code agents, software engineering
OpenAI Research team GPT series, Function Calling, Operator
Anthropic Research team Claude, Constitutional AI
DeepMind Research team Gemini, AlphaCode
Microsoft Research Research team AutoGen, TaskWeaver
Tsinghua KEG Jie Tang's team AgentBench, ChatGLM

References

  1. Weng, L. (2023). LLM Powered Autonomous Agents. lilianweng.github.io.
  2. Wei, J. et al. (2022). Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. NeurIPS 2022.
  3. Yao, S. et al. (2022). ReAct: Synergizing Reasoning and Acting in Language Models. ICLR 2023.
  4. Park, J.S. et al. (2023). Generative Agents: Interactive Simulacra of Human Behavior. UIST 2023.
  5. Shinn, N. et al. (2023). Reflexion: Language Agents with Verbal Reinforcement Learning. NeurIPS 2023.

评论 #